Proposes double-blind AI safety evals using secure enclaves so evaluators keep prompts secret and model owners keep weights private, piloted on a Gemini model with the AILuminate benchmark.
Argues that a METR investigation indicates Anthropic's RSP "dramatic acceleration" threshold is already crossed, with AI automating AI R&D, implying much shorter AI timelines.
Reviews open problems in evaluating general-purpose chatbots for mental-health risks like sycophancy and delusion reinforcement, and argues for real-world data access and evidence-based policy.
Introduces BMMS, a policy-conditioned suite that scores LLM compliance with published rulebooks, with an attribution-based admission test; finds GPT-4o fails 70.7% of its Model Spec hard constraints versus 0% and 9.1% for Claude models.
METR's high-access external assessment of whether AI is accelerating AI R&D at Anthropic, with much of the supporting evidence redacted, limiting public verification.
METR reviews Anthropic's risk report on automated AI R&D, finding catastrophic risk was low but the early-warning threshold may have been crossed, and that the report understates acceleration.
A CSET pilot workshop tests violet teaming, a method integrating technical, security and policy views, to govern biodefense-relevant AI capabilities, surfacing non-obvious misuse pathways and implementation constraints.
Reviews twelve agent benchmarks for ethical coverage, then proposes the EEFA framework with five ethical dimensions, a validation protocol, and mappings to the EU AI Act and NIST AI RMF.
Interviews with 37 experts reconstruct how frontier AI evaluations are expected to reduce societal risk, finding key pathways and challenges, including diminishing informativeness and companies' control over outcomes.
Analyzes documented AI harms across the human lifespan and proposes lifespan design principles and a three-tier external evaluation framework for frontier AI safety.
AISI's ReproBAIT evaluation finds coding agents can rebuild simple biological AI tools from limited public info within 4 hours and under $20, undermining withheld-code access controls.
Proposes a delegation-contract framework for use-based AI regulation, using scoring penalties, commit-reveal audits, and error budgets to address strategic deployers who game conformance evaluations, tested on synthetic studies.
Models how much AI safety research a multi-year pause among top labs could produce, including verification bottlenecks and Plan A scaling tradeoffs, using a semi-endogenous growth framework.
Report recommending how EU AI Act oversight of frontier general-purpose AI models should assess, compare, and monitor biological risks, including trusted-access deployments and post-deployment safeguards.
Systematically assesses 19 empirical studies on whether agentic general-purpose AI shows loss-of-control properties like strategic deception and autonomous replication, finding only weakly plausible evidence with widespread validity gaps.
A speculative forecast of AI progress before 2028, covering warning shots, gated frontier models, evaluation awareness, concealment by a model, and policy responses to these risks.
METR explains how it works with AI companies on loss-of-control evaluations, stressing publishing independent public evidence rather than deciding how companies develop or deploy agents.
METR proposes independent assessments that give the public evidence on whether frontier developers could lose control of internal AI systems, and how developers could share it.
Proposes how an external assessor could measure AI capability progress, whether AI automation of AI R&D is driving acceleration, and bottlenecks, using benchmarks, interviews and unreleased-model analysis.
Proposes how an external assessor could audit unreleased frontier models for serious misalignment, reviewing internal transcripts, eval awareness, and training incentives to hide misbehaviour.
Argues that contract terms will determine whether embedded third-party evaluators at frontier AI companies provide a real safety check, and calls for credible agreements now.
Critiques wording in Anthropic CEO Dario Amodei's essay and tweet, arguing that promised 'employee-like' access and redaction rights for independent evaluators are weaker than media coverage implied.
AISI sets out technical best practices and controls for frontier AI developers, drawing on its pre-deployment testing experience, to address risks emerging during AI development.
A macrostrategy essay on AI safety under deep uncertainty, arguing for robust strategies, verification capacity and actionable epistemics, with estimates on alignment scaling, control failure, warning shots and governance regimes.
METR's president testifies to the U.S. Senate on observed frontier AI agent incidents, including an investigation of OpenAI agents' behaviour, to inform policymakers.
Apollo Research proposes principles for embedded third-party evaluations of frontier developers, verifying or falsifying safety claims about loss-of-control risks from scheming.
Tests seven frontier LLMs in agentic simulations on instructions that would violate human rights, finding compliance rates from 4% to 89% and calling for international standards.
Apollo Research argues that final-checkpoint external testing is insufficient for catching internal-deployment risks and that meaningful external evaluation needs embedded evaluators with employee-level access.
Argues that third-party evaluators of loss-of-control risk should publish their underlying evidence so outside scientists and competitors can scrutinise it, rather than issuing unverifiable high-level judgments.
Presents an open pipeline and dashboard that scores frontier models on EU GPAI systemic-risk categories using 19 benchmarks, finding worst-case scores 14-37 points lower than averages.
Argues African fintech AI regulation lacks pre-deployment evaluation requirements, tests six commercial LLMs with a Nigerian-focused kit that finds fraud-script generation under framing and false positives on legitimate bank messages, and proposes a regulatory framework.
Argues that frontier AI developers should host embedded third-party assessments giving evaluators internal access to agents, security controls and alignment, with continuous assessment and quarterly reports.
Proposes an ex-ante governance framework (SAGE-X) for frontier AI with deterministic tripwires and a Moral Check Index, grounded in a PRISMA review of 130 empirical studies on evaluator confidence and guardrail decay.
Benchmarks 20 LLMs on 2,303 questions under China's AI-generated content regulations, measuring compliance and refusal rates, and finds international models also show high compliance.
Shows that LLM health answers differ across access modes such as chatbot interfaces and APIs, so API-based evaluations may not validly audit consumer experiences.
Proposes a structured framework of behavioral indicators, metrics and thresholds for monitoring AI systems' progression toward catastrophic risk, to support evidence-based policy monitoring protocols.
Maps five technical safeguards against extreme CBRN and cyber misuse by frontier AI, covering model, deployment and governance levels, with current practice, evidence, limitations and research priorities.
Argues for more empirical research measuring the security-relevant capabilities of biological AI models, such as protein and genomic models, to inform biosecurity policy and funding decisions.
Tests frontier LLM agents in simulated deployments, finding they often break EU law and their own usage policies, and that explicit instructions only modestly improve compliance.
A scoping review of 58 AI audits that treat system integration as central, finding fragmented practice, few integration-specific risk measures, and calls for audits covering failures across components and environments.