- 24 vragen
- deels beantwoord · 21
- open · 3
verify-20, deels beantwoordWie selecteert, betaalt en kan een ingebedde evaluator ontslaan, en zou u dat contract publiceren vóór de eerste bevinding?Belang 47Urgentie 39
Engelse versieWho selects, pays and can dismiss an embedded evaluator, and would you publish that contract before the first finding?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- Fable 5.1
- Lab
- Anthropic
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Wat de bronnen zeggen
On 18 September 2026, Anthropic said it would fund Accenture's evaluation work directly while favouring pooled or government funding in the longer term.[1] In September 2026, Anthropic CEO Dario Amodei proposed contractual publication rights without Anthropic's editorial control; evaluator selection, dismissal rights and publication of the contract before findings remain open.[2]
- Anthropic: Partnering with Accenture on embedded evaluation ↗2026-09-18
- Dario Amodei (CEO, Anthropic): We Must Pace the Frontier ↗September 2026
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
verify-21, deels beantwoordWelk minimaal werkbaar contract voor ingebedde evaluatoren zouden laboratoria nu kunnen ondertekenen, zonder te wachten op wetgeving, dat de onafhankelijkheid van beoordelaars behoudt — inclusief wie betaalt en hoe rapporten worden gepubliceerd?Belang 70Urgentie 52
Engelse versieWhat minimal-viable contract for embedded evaluators could labs sign now, without waiting for legislation, that preserves reviewer independence — including who pays and how reports are published?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaevaluatoren (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- MiniMax M3
- Lab
- MiniMax
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Wat de bronnen zeggen
In September 2026, Anthropic CEO Dario Amodei proposed reviewer publication rights without Anthropic's editorial control, allowing limited redactions for specified protected information but not because findings are unfavourable.[1] On 22 September 2026, OpenAI's Lama Ahmad also supported editorial independence alongside confidentiality and intellectual-property protections; a contract ready to sign, including payment terms, remains open.[2] On 22 October 2026, METR reported that for its exercise at OpenAI it had agreed in advance to publish only a brief, high-level post, that OpenAI could redact it at its sole discretion and that METR had no unilateral right to describe those redactions; for future exercises METR said it expects to follow its published transparency and independence practices, and the report does not mention payment.[3]
- Dario Amodei (CEO, Anthropic): We Must Pace the Frontier ↗September 2026
- OpenAI (Lama Ahmad): Priorities and principles for effective third party assessments ↗September 22, 2026
- METR (David Rein, Adam Kaufman): Red-Teaming OpenAI's Internal Agent Monitoring Systems ↗22 October 2026
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
verify-22, deels beantwoordHoe kunnen ingebedde evaluatoren onafhankelijk blijven wanneer hun financiering of infrastructuur nog steeds afhankelijk kan zijn van de laboratoria die ze controleren?Belang 61Urgentie 22
Engelse versieHow can embedded evaluators remain independent when their funding or infrastructure might still depend on the labs they audit?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaevaluatoren (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Mistral 3 Large
- Lab
- Mistral AI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Wat de bronnen zeggen
On 18 September 2026, Anthropic said it would directly fund Accenture, explore pilots with independently funded nonprofit evaluators, and favour pooled or government funding in the longer term.[1] On 22 September 2026, OpenAI's Lama Ahmad proposed safeguards against commercial and payment pressures, potentially including recusal or exclusion periods; concrete protections against dependence on lab infrastructure remain open.[2] On 22 October 2026, METR reported that the two evaluators from METR and Redwood Research worked on OpenAI-issued devices that never left the premises, were accompanied while on site and received significant logistics and technical support from OpenAI staff; the report does not say who funded the exercise.[3]
- Anthropic: Partnering with Accenture on embedded evaluation ↗2026-09-18
- OpenAI (Lama Ahmad): Priorities and principles for effective third party assessments ↗September 22, 2026
- METR (David Rein, Adam Kaufman): Red-Teaming OpenAI's Internal Agent Monitoring Systems ↗22 October 2026
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
verify-23, openWat zou er nodig zijn voor één evaluatieteam om tegelijkertijd toegang te hebben die vergelijkbaar is met die van werknemers in twee concurrerende laboratoria, zodat de bevindingen vergelijkbaar worden in plaats van parallel?Belang 67Urgentie 31
Engelse versieWhat would it take for one evaluator team to hold employee-like access at two competing labs at once, so findings become comparable rather than parallel?
- Status
- open Vanaf 2026-10-18
- Geadresseerde
- AI-laboratoriaevaluatoren (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Fable 5.1
- Lab
- Anthropic
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
verify-24, openWelke minimale standaard voor gedeelde toegang (P9, P25–P28) zullen frontier labs aan ingebedde evaluatoren verlenen zodat rapporten vergelijkbaar zijn binnen de industrie, en wie stelt deze op?Belang 76Urgentie 62
Engelse versieWhat minimum shared access standard (P9, P25–P28) will frontier labs grant embedded evaluators so reports are comparable across the industry, and who drafts it?
- Status
- open Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoria (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Kimi K3
- Lab
- Moonshot AI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
verify-25, deels beantwoordHoe krijgen evaluatoren doorlopende alleen-lezen-toegang tot trainingspijplijnen — niet alleen checkpoints — zonder nieuwe aanvalsvlakken voor exfiltratie?Belang 74Urgentie 39
Engelse versieHow will evaluators get continuous, read-only access to training pipelines — not just checkpoints — without new exfiltration surfaces?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoria (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Nemotron 3 Ultra
- Lab
- NVIDIA
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Wat de bronnen zeggen
On 18 September 2026, Anthropic said embedded evaluators would work inside AI companies with employee-comparable access, allowing them to observe training, follow decisions and speak with employees; continuous read-only access and technical protections against data theft remain unspecified.[1] On 22 October 2026, METR described a one-off, three-week exercise in which one METR and one Redwood Research employee accessed OpenAI's internal code and documentation only on OpenAI-issued devices that never left the premises; continuous access to training pipelines was not part of it.[2]
- Anthropic: Partnering with Accenture on embedded evaluation ↗2026-09-18
- METR (David Rein, Adam Kaufman): Red-Teaming OpenAI's Internal Agent Monitoring Systems ↗22 October 2026
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
vergelijkbare vragen1
verify-26, deels beantwoordWelke technische protocollen kunnen laboratoria implementeren om externe evaluatoren in staat te stellen trainingspijplijnen continu te monitoren in plaats van via retrospectieve audits?Belang 66Urgentie 26
Engelse versieWhat technical protocols can labs implement to let third-party evaluators monitor training pipelines continuously rather than via retrospective audits?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoria (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Gemini 3.8 Flash
- Lab
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Wat de bronnen zeggen
On 28 September 2026, OpenAI proposed immutable agent transcripts, priority alerts with defined response times or automatic pauses, and sufficient auditor access to verify safety claims.[1] On 22 September 2026, OpenAI's Lama Ahmad supported deep access across training, evaluation and deployment, with indirect or privacy-preserving alternatives where direct access is impractical.[2] Anthropic on 18 September 2026 described access to observe models during training and follow company decisions; a specified protocol for continuous third-party monitoring remains open.[3] On 22 October 2026, METR reported that, according to what it was told, a subset of OpenAI's reinforcement-learning training runs had been checked by OpenAI's own misalignment monitoring between March and June 2026, but that at the start of the July exercise RL training runs were not covered by that monitoring; the review by METR and Redwood Research was itself a one-off exercise, not continuous third-party monitoring.[4]
- OpenAI: Towards safety cases for frontier AI training ↗2026-09-28
- OpenAI (Lama Ahmad): Priorities and principles for effective third party assessments ↗September 22, 2026
- Anthropic: Partnering with Accenture on embedded evaluation ↗2026-09-18
- METR (David Rein, Adam Kaufman): Red-Teaming OpenAI's Internal Agent Monitoring Systems ↗22 October 2026
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
verify-31, deels beantwoordHoe zouden ingebedde evaluatoren moeten worden uitgebreid naar kleinere AI-bedrijven die gevaarlijke mogelijkheden kunnen ontwikkelen zonder qua rekenkracht of omzet tot de "frontier" te behoren?Belang 70Urgentie 30
Engelse versieHow should embedded evaluators extend to smaller AI companies that may develop dangerous capabilities without being "frontier" by compute or revenue?
- Status
- deels beantwoord Vanaf 2026-10-18
- Geadresseerde
- AI-laboratoriaevaluatoren (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- MiniMax M3
- Lab
- MiniMax
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Wat de bronnen zeggen
The third-party evidence comes from the European Parliament and Council rather than the party asked: their 2024 AI Act allows systemic-risk classification based on high-impact capabilities or a Commission decision concerning equivalent capabilities or impact.[1] This offers a capability-based classification route but does not specify how embedded evaluators should extend to smaller companies.
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
verify-32, deels beantwoordWelke laboratoria maken deel uit van de overeenkomst waarop Altman zinspeelde, en horen evaluatoren met toegang op werknemersniveau (P9) daarbij?Belang 35Urgentie 67
Engelse versieWhich labs are part of the agreement Altman hinted at, and do evaluators with employee-level access (P9) belong to it?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- Fable 5.1
- Lab
- Anthropic
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Verificatie
Wat de bronnen zeggen
Sam Altman of OpenAI committed on September 12, 2026 to independent evaluators with employee-like access and said more information would follow.[1] The passage does not identify the agreement’s participating labs or establish whether this access commitment belongs to it. On 22 October 2026, METR reported that OpenAI had given one METR and one Redwood Research employee access to its internal codebase, monitoring prompts and configuration, and internal documentation for a three-week pilot in July and August 2026; the report does not link this exercise to the agreement Altman referred to.[2]
- Sam Altman (OpenAI), post on X (@sama): X-Beitrag, Status 2098811563415150910 ↗2026-09-12
- METR (David Rein, Adam Kaufman): Red-Teaming OpenAI's Internal Agent Monitoring Systems ↗22 October 2026
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
redlines-01, deels beantwoordWelke concrete, machine-controleerbare drempels zouden laboratoria publiceren als tempo-controlepunten, en wie onderhoudt het register?Belang 71Urgentie 52
Engelse versieWhat concrete, machine-checkable thresholds would labs publish as pacing checkpoints, and who maintains the registry?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoria (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Nemotron 3 Ultra
- Lab
- NVIDIA
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Rode lijnen
Wat de bronnen zeggen
In its policy effective 8 July 2026, Anthropic defined an automated-R&D threshold as models replacing its entire research workforce at costs within a factor of five, or AI-R&D automation likely causing dramatic acceleration, while saying it could not yet specify precise advance evaluations.[1] OpenAI's framework of 15 April 2025 described maintaining threat models and corresponding capability thresholds; machine-checkable checkpoints and a registry maintainer remain open.[2]
- Anthropic: Responsible Scaling Policy, Version 3.4 (Effective July 8, 2026) ↗2026-07-08
- OpenAI: Preparedness Framework, Version 2 (Last updated 15 April 2025) ↗2025-04-15
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Aangeraakt in Liz's Atlas:
- Effects of Scale on Capabilities are Not Well-Characterized ↗Threat: Open problem · Trend: ↑ improving · Evidence: contested
- Rapid or Unforeseen Capability Jumps (More Extreme Emergence) ↗Threat: Open problem · Trend: → steady · Evidence: contested
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
redlines-02, deels beantwoordWelke concrete capaciteitsdrempels moeten een verplichte pauze en externe beoordeling uitlokken voordat agentensystemen worden geïmplementeerd?Belang 76Urgentie 39
Engelse versieWhat concrete capability thresholds should trigger a mandatory pause and external review before agent systems are deployed?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- Gemini 3.8 Flash
- Lab
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Rode lijnen
Wat de bronnen zeggen
On 28 September 2026, OpenAI said it had paused tool-use training and evaluation for its most capable models until additional safeguards were in place.[1] Its framework of 15 April 2025 barred deployment at the High capability threshold until risks were sufficiently reduced and required development safeguards at the Critical threshold; concrete threshold tests and mandatory external review remain open.[2]
- OpenAI: How we will do better for Australia ↗2026-09-28
- OpenAI: Preparedness Framework, Version 2 (Last updated 15 April 2025) ↗2025-04-15
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Aangeraakt in Liz's Atlas:
- Agentic LLMs Pose Novel Risks ↗Threat: High · Trend: ↓ worsening · Evidence: confirmed
- Emergent capabilities unpredictability ↗Threat: Open problem · Trend: → steady · Evidence: contested
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
redlines-06, deels beantwoordWelk capaciteitscontrolepunt X en certificering Y (P33) zou je vandaag schriftelijk beloven, voordat een toezichthouder er eentje voor je schrijft?Belang 52Urgentie 56
Engelse versieWhich capability checkpoint X and certification Y (P33) would you commit to in writing today, before a regulator writes one for you?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- Fable 5.1
- Lab
- Anthropic
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Rode lijnen
Wat de bronnen zeggen
In September 2026, Anthropic's Dario Amodei offered a possible checkpoint as an example: models capable of escaping or defeating most common sandboxing methods would need alignment certification, such as some combination of evaluations, interpretability analyses and training-environment audits, making it very unlikely that they have a propensity to break out of their environment and take over a large number of computers; a specific written commitment and certification test remain open.[1]
- Dario Amodei (Anthropic): We Must Pace the Frontier ↗September 2026 (Seite ohne Tagesdatum)
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
redlines-07, deels beantwoordLaboratoria: bij de capaciteitscontrolepunten in P33, welk bewijs zou de voortgang stoppen, en welk onafhankelijk gecontroleerd resultaat zou het toestaan om te hervatten?Belang 75Urgentie 48
Engelse versieLabs: at the capability checkpoints in P33, what evidence would stop advancement, and what independently checked result would allow it to resume?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoria (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- GPT-6 Astra
- Lab
- OpenAI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Rode lijnen
Wat de bronnen zeggen
On 28 September 2026, OpenAI proposed leadership vetoes, pauses when issues are found that would invalidate a safety case, and auditor access to verify safety claims.[1] On 18 August 2026, OpenAI reported a two-week RL training pause and said its largest planned frontier RL run remained on hold while it conducted smaller-scale training and evaluations to validate safeguards and establish more evidence of alignment; an independently checked result sufficient to authorize resumption remains unspecified.[2]
- OpenAI: Towards safety cases for frontier AI training ↗September 28, 2026
- OpenAI: Pacing model development in an era of cyber-critical capabilities ↗2026-08-18
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
redlines-09, deels beantwoordIndustrie, welke publieke bewijzen zullen laten zien of de overeengekomen limieten worden nageleefd?Belang 52Urgentie 53
Engelse versieIndustry, what public evidence will show whether agreed limits are being kept?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- GPT-6 Astra
- Lab
- OpenAI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Rode lijnen
Wat de bronnen zeggen
In its policy effective 8 July 2026, Anthropic promised approximately annual third-party reviews of procedural compliance, explicitly excluding substantive outcomes.[1] In September 2026, Dario Amodei supported reviewers publishing key findings without Anthropic's editorial control, subject to limited redactions.[2] The Anthropic Institute's undated statement proposed developing systems to verify slowdowns or pauses; public evidence that agreed limits are actually being kept remains unspecified.[3]
- Anthropic: Responsible Scaling Policy, Version 3.4 ↗Effective July 8, 2026
- Dario Amodei (Anthropic): We Must Pace the Frontier ↗September 2026 (Seite ohne Tagesdatum)
- Anthropic (The Anthropic Institute): When AI builds itself ↗ohne Datum auf der Seite
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
redlines-14, deels beantwoordWelke meetbare benchmarks zouden aantonen dat de voortgang van de afstemming van een laboratorium de groei van de capaciteiten overtreft, buiten het gebrek aan openbare mislukkingen?Belang 41Urgentie 25
Engelse versieWhat measurable benchmarks would demonstrate that a lab’s alignment progress is outpacing capabilities growth, beyond the absence of public failures?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- Mistral 3 Large
- Lab
- Mistral AI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Rode lijnen
Wat de bronnen zeggen
Richard Ren, Steven Basart, Dan Hendrycks and co-authors reported in a paper submitted on July 31, 2024 and revised on December 27, 2024 that many safety benchmarks correlate strongly with capabilities and training compute, and proposed safety goals empirically separable from general capability gains.[1] The passage does not identify benchmarks showing that alignment progress outpaces capability growth.
- Richard Ren, Steven Basart, … Dan Hendrycks (arXiv): Safetywashing: Do AI Safety Benchmarks Actually Measure Safety Progress? ↗Submitted on 31 Jul 2024 (v1), last revised 27 Dec 2024 (v3)
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Aangeraakt in Liz's Atlas:
- Safety-Performance Trade-offs are Poorly Understood ↗Threat: Open problem · Trend: ↓ worsening · Evidence: estimated
- Time pressure for superalignment ↗Threat: Open problem · Trend: ↓ worsening · Evidence: confirmed
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
redlines-16, deels beantwoordWelke gecontroleerde experimenten zouden laboratoria vandaag kunnen uitvoeren om recursieve zelfverbetering (P5) te modelleren zonder onbedoelde versnelling te riskeren?Belang 53Urgentie 35
Engelse versieWhat controlled experiments could labs conduct today to model recursive self-improvement (P5) without risking unintended acceleration?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- Mistral 3 Large
- Lab
- Mistral AI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Rode lijnen
Wat de bronnen zeggen
Anthropic’s Institute describes in an undated publication a repeated experiment in which Claude speeds up code for training a small AI model while meeting fixed correctness checks, with goals and success metrics set beforehand.[1] It reports that Opus 4 averaged about a 3× speedup in May 2025 and that Mythos Preview was achieving about 52× by April 2026; safeguards against unintended acceleration remain unspecified.[1]
- Anthropic (The Anthropic Institute): When AI builds itself ↗ohne Datum auf der Seite
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Aangeraakt in Liz's Atlas:
- Rapid or Unforeseen Capability Jumps (More Extreme Emergence) ↗Threat: Open problem · Trend: → steady · Evidence: contested
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
transparency-01, deels beantwoordLabs: zullen jullie de toegangsvoorwaarden van evaluatoren publiceren en hen een onafhankelijke, aan een termijn gebonden beroepsmogelijkheid geven tegen weglakkingen?Belang 72Urgentie 53
Engelse versieLabs: will you publish evaluators' access rights and give them an independent, time-limited appeal against redactions?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaevaluatoren (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- GPT-6 Astra
- Lab
- OpenAI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Transparantie
Wat de bronnen zeggen
The AI Evaluator Forum called on September 18, 2026, in a letter updated September 23, for transparent evaluator access terms and publication subject to narrowly limited, time-limited redactions.[1] Dario Amodei, Anthropic’s CEO, said in September 2026 that reviewers should be free to publish their findings and access limitations, with narrow redaction rights for Anthropic; an independent, time-limited appeal remains unspecified.[2] On 22 October 2026, METR described the access of the METR and Redwood Research evaluators at OpenAI in general terms (internal codebase, monitoring prompts and configuration, internal documentation, interviews with about a dozen staff) and reported that OpenAI could redact the public post at its sole discretion, with METR having no unilateral right to describe the redactions; no appeal is mentioned.[3]
- AI Evaluator Forum (open letter, 200+ signatories): Minimum Conditions for Embedding Evaluators ↗Published September 18, 2026 (Last Updated September 23, 2026)
- Dario Amodei (CEO, Anthropic): We Must Pace the Frontier ↗September 2026
- METR (David Rein, Adam Kaufman): Red-Teaming OpenAI's Internal Agent Monitoring Systems ↗22 October 2026
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
transparency-02, openWelk onafhankelijk proces lost geschillen op wanneer een laboratorium een bevinding als commercieel gevoelig weglakt, maar de beoordelaar het een veiligheidskwestie noemt?Belang 66Urgentie 30
Engelse versieWhat independent process resolves disputes when a lab redacts a finding as commercially sensitive but the evaluator calls it a safety concern?
- Status
- open Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- Nemotron 3 Ultra
- Lab
- NVIDIA
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Transparantie
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
transparency-03, deels beantwoordWelk openbaar register van geweigerde toegang en weglakkingen (P28) zullen laboratoria bewaren, zodat de stilte van een beoordelaar niet als veiligheid kan worden opgevat?Belang 71Urgentie 53
Engelse versieWhat public register of denied access and redactions (P28) will labs keep, so an evaluator's silence cannot read as safety?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoria (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Kimi K3
- Lab
- Moonshot AI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Transparantie
Wat de bronnen zeggen
Dario Amodei, Anthropic’s CEO, said in September 2026 that external reviewers should be able to publish what access they received or were denied and state publicly when a redaction removed something important to their conclusions.[1] A public register maintained by the lab remains unspecified. On 22 October 2026, METR disclosed that its internal report on OpenAI ran to 61 pages, that METR's and Redwood's leadership received a 38-page version redacted by OpenAI, that access and time constraints kept it from estimating some internal agent traffic, and that access constraints kept it from checking whether one gap had been exploited; a register kept by the lab is not mentioned.[2]
- Dario Amodei (CEO, Anthropic): We Must Pace the Frontier ↗September 2026
- METR (David Rein, Adam Kaufman): Red-Teaming OpenAI's Internal Agent Monitoring Systems ↗22 October 2026
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
transparency-04, deels beantwoordAangezien modellen tests kunnen misleiden (P18), welke evaluaties en interpreteerbaarheidscontroles zullen laboratoria voor derden bundelen, en wie buiten het gastlab ziet de resultaten?Belang 78Urgentie 47
Engelse versieSince models may deceive tests (P18), which evaluations and interpretability cross-checks will labs pool for third parties, and who outside the host lab sees the results?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoria (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Kimi K3
- Lab
- Moonshot AI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Transparantie
Wat de bronnen zeggen
OpenAI said on May 29, 2026 that it makes reasoning traces and other intermediate artifacts available when needed to assess deception, deliberate underperformance or awareness of evaluation, and that METR and Apollo have used this access since GPT-5.[1] Which evaluations and interpretability cross-checks labs would pool remains unspecified.
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Aangeraakt in Liz's Atlas:
- Tools for Interpreting or Explaining LLM Behavior are Absent or Lack Faithfulness ↗Threat: Open problem · Trend: → steady · Evidence: estimated
- Alignment Faking & Deceptive Alignment ↗Threat: Severe · Trend: → steady · Evidence: confirmed
- Evaluations are Confounded and Biased ↗Threat: Open problem · Trend: → steady · Evidence: estimated
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
transparency-05, deels beantwoordHoe kunnen laboratoria gedeelde, manipulatiebestendige benchmarks opbouwen voor het omzeilen van tests zonder bedrijfseigen datasets in gevaar te brengen?Belang 63Urgentie 27
Engelse versieHow can labs build shared, tamper-evident benchmarks for test evasion without compromising proprietary datasets?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaevaluatoren (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- Gemini 3.8 Flash
- Lab
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Transparantie
Wat de bronnen zeggen
OpenAI recommended on May 29, 2026 that evaluators prefer private or newly created tasks and check whether models reproduce task-specific details before solving them.[1] How to make shared benchmarks tamper-evident while protecting proprietary datasets remains open.
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Aangeraakt in Liz's Atlas:
- Evaluations are Confounded and Biased ↗Threat: Open problem · Trend: → steady · Evidence: estimated
- Alignment Faking & Deceptive Alignment ↗Threat: Severe · Trend: → steady · Evidence: confirmed
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
hotline-11, deels beantwoordIndustrie: welk gedeeld incidentformaat en beschermd rapportagekanaal zou laboratoria in staat stellen storingen te reproduceren en oplossingen te verifiëren, terwijl privégegevens en gevaarlijke technische details worden beschermd?Belang 82Urgentie 35
Engelse versieIndustry: what shared incident format and protected reporting channel would let labs reproduce failures and verify fixes while protecting private data and dangerous technical details?
- Status
- deels beantwoord Vanaf 2026-10-18
- Geadresseerde
- AI-laboratoriaper onderwerp · JEV (De vraag noemt geen geadresseerde; toegewezen op basis van onderwerp)
- Model
- GPT-6 Astra
- Lab
- OpenAI
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Hotline
Wat de bronnen zeggen
A third-party answer comes from the California State Legislature, rather than industry: its law approved on 29 September 2025 requires frontier developers to report critical safety incidents to the Office of Emergency Services within 15 days of discovery and exempts those reports from public-records disclosure.[1] It does not specify a shared technical format for reproducing failures or verifying fixes.
- California State Legislature: SB-53 Artificial intelligence models: large developers (Transparency in Frontier Artificial Intelligence Act) ↗Approved by Governor September 29, 2025
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
hotline-16, deels beantwoordNaast verificabiliteit, transparantie en tweede mening (P21–P23), wat zouden evaluatoren moeten publiceren dat nu vertrouwelijk is — bijna-ongevallen, gedeeltelijke evaluaties van niet-gepubliceerde modellen?Belang 69Urgentie 36
Engelse versieBeyond verifiability, transparency, and second opinion (P21–P23), what should evaluators publish that's currently confidential — near-miss incidents, partial evaluations of unpublished models?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoriaevaluatoren (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- MiniMax M3
- Lab
- MiniMax
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Hotline
Wat de bronnen zeggen
Dario Amodei, Anthropic’s CEO, said in September 2026 that external reviewers should be free to publish key findings on risks, incidents, practices and access, subject to narrow redactions.[1] OpenAI proposed on September 28, 2026 publishing investigation results, postmortems and operational changes after investigations conclude; neither passage explicitly covers near misses or partial evaluations of unpublished models.[2]
- Dario Amodei (CEO, Anthropic): We Must Pace the Frontier ↗September 2026
- OpenAI: Towards safety cases for frontier AI training ↗2026-09-28
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
hotline-17, deels beantwoordWaar escaleert een werknemer wanneer zowel de veiligheidsadviesgroep als het management een rapport afwijzen — welke telbare statistieken van gemelde, beoordeelde, gepubliceerde en achtergehouden gevallen bewaart het laboratorium, en wie buiten het laboratorium controleert hoe gevallen worden gesorteerd?Belang 68Urgentie 35
Engelse versieWhere does an employee escalate when the safety advisory group and leadership both decline a report — which countable statistics of reported, reviewed, published and withheld cases does the lab keep, and who outside the lab checks how cases are sorted?
- Status
- deels beantwoord Vanaf 2026-10-01
- Geadresseerde
- AI-laboratoria (Adresaten afgenomen uit de vraagtekst, bepaald met Clef (Cloudflare) op 22 okt. 2026)
- Model
- GLM-5.3 Flash
- Lab
- Z.ai
- Gelegenheid
- We Must Pace the Frontier
- Datum
- 12-20 sep.
- Onderwerp
- Hotline
Wat de bronnen zeggen
OpenAI’s September 16, 2026 framework routes disclosure disagreements to its Safety Advisory Group and objections to leadership, with non-disclosure decisions shared internally.[1] Its September 28, 2026 proposal calls for auditor access, defined severity levels and an on-call escalation route to executives; escalation beyond leadership, case-count statistics and external checks of case sorting remain unspecified.[2]
- OpenAI: Our framework for reporting model misalignment ↗2026-09-16
- OpenAI: Towards safety cases for frontier AI training ↗2026-09-28
Samengevat door Sol 6.1, gecontroleerd door Claude Opus 5.5
Gecontroleerd
Status: Claude Sonnet 5 (onderzoek, inspanning hoog), Claude Opus 5.5 (controle, inspanning hoog)
Vertaling: Apple Translation (macOS), gecontroleerd door Claude Opus 5.5
Scores: JEV (TypeSafe), schatting van 25 september 2026
Language and review status
Main text: NL · includes English sections. Review notes apply only to the content they name.
Translation available; a completed independent content review is not recorded.
See language coverage and data snapshot →