Highlights from Garrison Lovely's "Obsolete"
A review of Garrison Lovely's book on AI's labor-displacement risks, arguing for pause advocacy, liability for uninsurable risks, and stronger governance against lobbying and unchecked deployment.
Research / Sources
LessWrong
AI research and discussion on LessWrong. www.lesswrong.com
AI (2,246)Interpretability (ML & AI) (225)AI Control (141)AI Governance (101)AI Evaluations (91)Chain-of-Thought Alignment (46)AI Risk (33)AI Safety (31)AI Takeoff (29)AI-Assisted Alignment (22)AI Rights / Welfare (20)AI Alignment Fieldbuilding (19)AI Timelines (19)Constitutional AI (15)AI Psychology (14)Dealmaking (AI) (14)Regulation and AI Risk (14)AI Benchmarking (13)Jailbreaking (AIs) (13)AI Alignment Intro Materials (11)Alignment Pretraining (11)Slowing Down AI (10)Aligned AI Proposals (9)Debate (AI safety technique) (9)AI Capabilities (8)AI Development Pause (8)OpenAI (8)Aligned AI Role-Model Fiction (7)Human-AI Safety (7)Threat Models (AI) (7)AI Oversight (6)AI Sentience (6)AI Safety Cases (5)AI "Agent" Scaffolds (4)AI arms race (4)AI Safety Camp (4)AI Safety Public Materials (4)Open Source AI (4)Transformative AI (4)Adversarial Examples (AI) (3)Adversarial Training (3)AI Alignment (3)AI Boxing (Containment) (3)AI Persuasion (3)AI Risk Concrete Stories (3)AI Risk Skepticism (3)AI Success Models (3)Guaranteed Safe AI (3)Sandbagging (AI) (3)AI Consciousness (2)AI Misuse (2)AI Racing (2)AI Spiralism / AI Psychosis (2)Frontier AI Companies (2)Grants & Fundraising Opportunities (2)PauseAI (2)Prosaic Alignment (2)Public Reactions to AI (2)Whole Brain Emulation (2)AI Auditing (1)AI Ethics (1)AI Lab Self-Immolation (1)AI Questions Open Threads (1)AI Robustness (1)AI Services (CAIS) (1)AI-Fizzle (1)AIXI (1)Blackmail / Extortion (1)Brainstorming (1)Power Seeking (AI) (1)
Focus All
Topic All
Date Any time
D̶o̶n̶’̶t̵⠀Can’t Look Up
Argues that AI development is hard to pause because of a visibility problem (unclear risks) and a structural problem (state incentives in anarchic international order), and proposes what coordination would need.
1000 Years of AI Safety
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.
"Holy shit" graphs and how to get more of them
Argues that labs should publish auditable capability and deployment metrics, such as internal spending, revenue and AI R&D automation trends, so policymakers can track frontier risks like loss of control and gradual disempowerment.
How far is ZKP from compute verification in high-stakes environments?
Estimates how far zero-knowledge proofs of LLM inference are from cost-competitive with auditing for verifying AI compute in international agreements, measuring roughly 1,700-3,800x proving overhead and hash-function compatibility issues.
China AI Safety Regulation Monthly Monitor, Issue 1 (September 2026)
Analyses China's AI safety regulation, including the non-binding Framework 3.0 on loss-of-control risks, CAC model filing gaps, content-focused enforcement, and a court opinion on developer liability.
Do AI models assist with human rights violations?
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.
We need a better theory of polarization, because it's failing to predict the AI debate
Argues that standard polarization theory fails to predict US politics on AI, noting bipartisan concern about AI x-risk and data centers, and draws strategic implications for AI regulation.
Stopgap Measures to Address Immediate AI Security Threats
ControlAI proposes three stopgap policies against immediate superintelligence-related risks: securing weapons-grade AI against theft, criminal liability for AI containment leaks, and government kill-switches.
Pacing the Frontier: A Framework & Research Agenda
Proposes pacing AI development, meaning interventions that moderate its pace, as a dedicated research field, covering incentives, regulatory and technical tools, and the full lifecycle of interventions.
Open Compute Verification Observatory as a First Metrological Layer for Technical AI Governance
Proposes a shared, public-cluster observatory of GPU telemetry and other physical signals to verify AI training runs, as a first measurement layer for frontier AI governance.
Let's talk about the AI coordination problem
Argues that coordination among AI developers to avoid building dangerous AI deserves serious, concrete discussion like a real geopolitical negotiation, and that leaders should explain why they consider it infeasible.
How the American Executive Could Control AI Companies
Analyses US federal statutes and executive powers that let the president unilaterally influence or control frontier AI companies, and the limits imposed by Congress and courts.
Frontier agents don't comply with standards, even when instructed to
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.
On using crises to shift political will for AI
Argues that AI safety incidents and their felt urgency can shift political will for frontier AI regulation, drawing on precedents like nuclear treaties, Montreal Protocol, and COVID responses, and urges advocates to exploit them.
AI-amplified democratic backsliding: an exploration
Builds a pilot index scoring 27 countries' structural vulnerability to AI-amplified democratic backsliding across five drivers, finding liberal democracies most resilient and polarization a key weakness.
Traffic Shaping for Workload Classification
Proposes a network-based verification design, traffic shaping plus randomized routing, that estimates it would make covert frontier training runs economically infeasible and so support treaty compliance.
Is a pause enforceable? New paper out!
Proposes a taxonomy for verifying and enforcing international AI pauses via compute monitoring, estimating that pauses are enforceable for training runs above roughly 10k H100-equivalents absent concealment.
AI Rights Aren't Safety-Neutral: A Quick Follow-Up to the Consciousness Cluster
Prompting or fine-tuning models to claim or deny legal rights shifts power-seeking and corrigibility benchmarks, suggesting AI rights framing is not safety-neutral and should inform legislation.
Estimating LLM Training FLOPs on the Nvidia Jetson Orin Nano
Proposes and tests a side-channel method that estimates LLM training FLOPs from Jetson GPU power and memory readings, to help verify compute thresholds in AI treaties, reaching about 4.5% mean error.
Don't normalize a permanent underclass (even a rich one)
Argues that post-AGI cosmic resources should not be allocated by pre-AGI wealth, since tradable claims would create a permanent underclass and race incentives, and proposes inalienability and staged allocation as alternatives.
AI Safety Policy Needs to train Legal Practitioners
An essay arguing that AI safety policy needs legal practitioners who can implement regulation, using GDPR's Data Protection Officer failures as a warning and describing efforts to train such people in AI safety.
When is misalignment just a bug?
Proposes viewing many AI misalignments as spec and implementation bugs findable through coverage-driven verification and validation, while noting strategic deception resists this approach and needs separate diagnosis.
We need 3rd party Training-Run Evaluations
Argues that third-party evaluations of frontier training runs, covering checkpoints, data and process decisions, are needed to detect scheming that final-checkpoint testing may miss, and proposes a phased access plan.
Model access for third-parties — it's a big deal!
Argues external safety researchers and auditors need model access comparable to frontier-lab insiders, analysing access gaps from non-release, deployment lag, safeguards, costs and elicitation.
A system overview for near-term, low-trust AI compute verification (deprecated, update coming soon)
Proposes a low-trust reference architecture for verifying AI compute, using network taps, air-gapped or ZKP-based evaluation, and workload evidence capture to check treaty compliance.
The Invisible Side of AI Governance
A LessWrong strategy post arguing that much impactful AI governance work is invisible insider policy-making in executive branches and multilateral forums, and that the safety community underinvests in it.
The Three Filters: Why Almost Every Plan to Survive ASI Fails Miserably
Argues that common theories of change for surviving ASI fail under race dynamics, covering nuclear war, misaligned AI, and nightmare singletons, and that safety work can boost capabilities, so international coordination is needed.
How valuable are weak AI safety regulations?
Argues that weak AI regulations such as GPU export controls, mandatory safety testing and whistleblower protections can modestly reduce extinction risk, weighing their routes to impact against downsides.
Coverage-driven alignment - What ‘Teaching Claude Why’ can borrow from AV verification
Proposes borrowing coverage-driven verification from autonomous-vehicle testing to systematically train and evaluate alignment, using explicit coverage maps over scenario dimensions for long-horizon AI agents; largely a conceptual proposal without empirical results.
What if Anthropic unilaterally paused capabilities development right now?
A LessWrong opinion piece arguing that Anthropic should unilaterally pause frontier capabilities development for safety, analysing how this would affect other labs, governments, and Anthropic's IPO.
Compute Verification on Short Timelines
Argues that compute verification is key to AI coordination, that die-level verification is too slow for short timelines, and that board- and firmware-level verification for Rubin and Feynman chips is more tractable and must start now.
No frontier model has acceptable levels of compliance with the EU AI Act and privacy legislation.
Introduces LARA, an agentic simulation tool that tests frontier models for violating EU AI Act and GDPR provisions, finding most models break the law in many scenarios, with the best at 54% compliance.
Engineering a Safer World: Risk Modelling — and Safety Engineering? — for AI Loss of Control
Applies Leveson's systems-safety thinking to AI loss of control, modelling how breaches at each oversight level can escalate into vicious cycles and how AI could also strengthen governance.
The case for fine-grained tracking of compute for AI
Argues that FLOP/s and bandwidth proxies no longer track AI compute well and proposes architecture-specific compute scaling models, to improve forecasts and compute governance thresholds.
Voters are surprisingly open to talking about AI risk
A LessWrong post arguing that US voters are unexpectedly open to discussing AI existential risk, and offering canvassing tactics for AI-safety-friendly candidates like Alex Bores.
When should an AI incident trigger an international response? Criteria for international escalation and implications for the design of AI incident frameworks
Proposes an eight-criterion framework for deciding when an AI incident should trigger international escalation, stress-tests it against real incidents, and identifies design gaps in the EU AI Act and SB 53.
Could Frontier AI Researchers Collectively Slow the Race? A Conditional Pledge Mechanism
Proposes a conditional pledge mechanism in which frontier AI researchers would collectively step away from capability work if enough peers across labs sign, and seeks feedback on feasibility and barriers.
Dawn of the "national security" tier of AI
Speculative commentary arguing that a US government 'national security' tier of frontier AI, with pre-release vetting and public-private partnerships, is emerging after Claude Mythos's hacking capabilities, and naming likely political players.
Only Politics Can Prevent Extinction*
An opinion post arguing that electoral politics and a pro-AI-safety political movement are necessary to get strict AI regulation enacted, citing historical cases of popular policies stalled by legislative inertia.