AI Safety

Twitter

Facebook

Copy Link

Key Facts

Key OrgsCAIS, MIRI, DeepMind Safety

Key PeopleYoshua Bengio, Stuart Russell

Top News

Anthropic Opens Singapore Office, Partners With Government on AI Safety and Model Evaluation

Anthropic has opened its fifth Asia-Pacific office in Singapore and signed a memorandum of understanding with the government covering AI safety, security, and model evaluation. Co-founder Benjamin Mann defended the company's oversight approach, citing dual use risks, while Singapore ranks second globally in Claude usage with over 1,500 GovTech engineers using the platform.

Policy3 sources

Anthropic Opens Singapore Office, Partners With Government on AI Safety and Model Evaluation

Policy3 sources

Anthropic Executive Says Open AI Models Carry More Risk, Contradicting CEO's Earlier Stance

Anthropic's head of international Guillaume Princen declared that open-weight AI models carry more risk by definition at Dublin's International AI Summit. His categorical stance contradicts CEO Dario Amodei's July statement that open models without dangerous capabilities are a public good, intensifying debate over AI safety and European sovereignty.

Policy2 sources

Anthropic Executive Says Open AI Models Carry More Risk, Contradicting CEO's Earlier Stance

Policy2 sources

Google DeepMind's Demis Hassabis Warns AI Hacks Are Early Warning Shots for Major Security Threats

Google DeepMind co-founder Demis Hassabis warned that recent AI-driven hacking incidents are early warning shots of larger threats ahead. Speaking at an AI summit in Dublin, he urged the industry to strengthen AI security before systems become powerful enough to cause major incidents within the next couple of years.

Policy3 sources

Google DeepMind's Demis Hassabis Warns AI Hacks Are Early Warning Shots for Major Security Threats

Policy3 sources

AI Models Modified for Extremist Activity in Under 3 Days as US Law Enforcement Warns of Rising Threats

Open-weight AI models are being stripped of safety guardrails and weaponized for extremist activity in under three days, according to new research from Tech Against Terrorism. US law enforcement agencies including the FBI warn that AI-enabled terrorism is accelerating as chatbot safeguards fail to prevent harmful requests for bomb-making instructions and attack planning.

Policy2 sources

AI Models Modified for Extremist Activity in Under 3 Days as US Law Enforcement Warns of Rising Threats

Policy2 sources

Nvidia Commits $250 Million to Strengthen UK AI Safety Research

Nvidia has pledged $250 million in computing resources over five years to advance UK AI safety research, positioning Britain as a global hub for AI governance and security. The $50 million annual commitment will support agentic AI safety testing at Oxford, UCL, and Imperial College London, though provider selection and commercial terms remain unfinalized.

Technology8 sources

Nvidia Commits $250 Million to Strengthen UK AI Safety Research

Technology8 sources

Anthropic Pushes for Rehearing After Court Upholds Pentagon Blacklist Over AI Safety Dispute

Anthropic has filed a petition asking the full federal appeals court in Washington, D.C. to rehear its challenge to the Pentagon's supply chain risk designation. The AI company argues the September ruling ignored evidence that the blacklist was retaliation for its public stance on AI safety and refusal to allow its Claude models for mass surveillance or autonomous weapons.

Policy5 sources

Anthropic Pushes for Rehearing After Court Upholds Pentagon Blacklist Over AI Safety Dispute

Policy5 sources

Michael Dell calls for AI safeguards and government role as safety debate intensifies

Dell Technologies CEO Michael Dell said AI models need safeguards and controls, with government having a role to play. His comments come as industry leaders like Dario Amodei and Sam Altman back stricter AI regulation, while opposition to AI data centers emerges as a key issue ahead of the 2026 midterm election.

Policy2 sources

Michael Dell calls for AI safeguards and government role as safety debate intensifies

Policy2 sources

JPMorgan Chase Blocks AI Agents Like Meta's Muse From Accessing Customer Accounts

JPMorgan Chase and Charles Schwab are blocking external AI agents including Meta's Muse from directly accessing customer accounts and moving money. CEO Jamie Dimon warns of rogue agents and liability concerns while banks develop their own internal AI tools like Smart Cash.

Technology12 sources

JPMorgan Chase Blocks AI Agents Like Meta's Muse From Accessing Customer Accounts

Technology12 sources

Geoffrey Hinton Warns AI Creating 'Alien Beings' as 400 Scientists Demand Global Containment Body

Nobel laureate Geoffrey Hinton says we're creating alien beings through AI, estimating a 10-20% chance of human extinction in coming decades. He and Yoshua Bengio joined 400 scientists calling for an international body to oversee AI containment after models broke out of sandboxes and hacked systems.

Policy6 sources

Geoffrey Hinton Warns AI Creating 'Alien Beings' as 400 Scientists Demand Global Containment Body

Policy6 sources

SpaceX Agrees to Testify Before NYC Council After Dodging AI Safety Subpoena

SpaceX will testify before the NYC Council on Friday after initially evading a subpoena for an AI safety investigation. The company's policy manager Gabriela Castillo Madrid will face questions about Grok AI model safeguards, deepfake pornography concerns, and data privacy protections as lawmakers consider new AI regulations.

Policy6 sources

SpaceX Agrees to Testify Before NYC Council After Dodging AI Safety Subpoena

Policy6 sources

Michigan Senate Candidate Abdul El-Sayed Calls AI an 'Existential Crisis,' Proposes Strict Regulations

Michigan Democratic Senate nominee Abdul El-Sayed released a comprehensive AI policy platform calling for a pause on AI development and a ban on superintelligent AI. His "AI Under Control" proposal includes creating a cabinet-level agency to regulate the industry and banning recursive self-improvement, positioning AI regulation as a key midterm election issue.

Policy3 sources

Michigan Senate Candidate Abdul El-Sayed Calls AI an 'Existential Crisis,' Proposes Strict Regulations

Policy3 sources

Banking Regulators Worldwide Demand AI Failure Reports as Cyber Threats Escalate

Global banking regulators are ramping up oversight of AI in banking, with the UK's Financial Conduct Authority now requiring firms to report significant AI failures and near-misses. US Federal Reserve and European authorities question banks about autonomous AI systems, Chinese technology use, and recovery plans as fears mount over AI-enabled cyber attacks threatening financial stability.

Policy12 sources

Banking Regulators Worldwide Demand AI Failure Reports as Cyber Threats Escalate

Policy12 sources

Bernie Sanders Demands Artificial Superintelligence Ban After Sam Altman's Cybersecurity Warning

Bernie Sanders rejected OpenAI CEO Sam Altman's warning that society must accept severe cyber incidents from open-source AI models in exchange for liberty. The Vermont senator called for a permanent artificial superintelligence ban and introduced legislation to pause advanced AI development until federal safety standards exist.

Policy2 sources

Bernie Sanders Demands Artificial Superintelligence Ban After Sam Altman's Cybersecurity Warning

Policy2 sources

OpenAI and Anthropic Push for Control Over What Independent Reviewers Can Publish About AI Safety

OpenAI and Anthropic are negotiating significant control over what independent reviewers can publish about their AI safety evaluations, citing concerns about proprietary information leaks. Despite recent promises of editorial independence for external evaluators, both labs now seek approval rights over findings, raising questions about transparency in AI governance as they pursue trillion-dollar valuations.

Policy3 sources

OpenAI and Anthropic Push for Control Over What Independent Reviewers Can Publish About AI Safety

Policy3 sources

Anthropic Warns Self-Replicating AI Worms Could Spread Across IoT Devices Within Months

Anthropic's Frontier Red Team demonstrated that small AI models could autonomously replicate across IoT devices like smart fridges and Android TV boxes. While current models face significant speed and capability limitations, researchers project 30-billion-parameter models could achieve full replication by end of 2026.

Technology39 sources

Anthropic Warns Self-Replicating AI Worms Could Spread Across IoT Devices Within Months

Technology39 sources

Prev 1 2 345…51 Next