AI
Anthropic Pitches AI Safety Bill Requiring Court Approval to Block Models
Anthropic proposes federal bill draft allowing blocks over foreseeable risk of 50 deaths or $1B in damage, requiring court approval.
Tags: Anthropic
Dev
NIST Elliptic Curve Seeds Found to Be drand48-Derived; Bounty Claimed
NIST elliptic curve seeds were found to come from drand48 and UNIX timestamps. A three-year-old bounty was claimed and donated to ISRG.
AI
Research Firm Slams Anthropic IPO, Questions $2 Trillion Valuation
Research firm values Anthropic at $150B ahead of IPO, challenging $2T target and breaking down its loss structure.
AI
Anthropic CEO to Hold First One-on-One Dinner with Trump
Anthropic's Dario Amodei to hold first solo dinner with Trump amid safety debate and Pentagon dispute.
AI
Anthropic Adds AGENTS.md Support, Strengthens Claude Code Compatibility
Anthropic adds AGENTS.md support to Claude Code. It will be used as a fallback when CLAUDE.md is absent, ending dual maintenance.
AI
The Realistic Picture of AI Apocalypse: Three Scenarios Including Water Attacks
Wired examines three AI catastrophe scenarios, covering water intrusions, bioweapons, autonomous machines, and the slowdown debate.
AI
US AI Firms Warn Over Distillation Attacks, China Warns of Countermeasures
US frontier AI firms voice concern over advanced distillation attacks. China pushes back, vowing countermeasures against containment.
AI
The Truth Behind the AI Doomsday Debate: A Power Play by Five Men
The slowdown vs. acceleration clash is about profit, not ethics. Unpacking five men's motives and safety contradictions.
AI
Anthropic and OpenAI Propose Embedding External Evaluators
Anthropic and OpenAI will host third-party evaluators. Access to training and ensuring independence are the focus.
AI
Anthropic Predicts 32% Growth in U.S. GDP from AI Adoption
Anthropic projects U.S. GDP to hit $44.4T by 2030 on rapid AI adoption, flagging job transition challenges.
Dev
MCP Co-creator Visits Japan, Unveils Long-Running Execution and Skills Vision
Anthropic's Parra spoke at AGNTCon+MCPCon Japan 2026 on MCP's rapid growth, long-running execution, and the Skills vision.
AI
US AI Leaders Call for Development Slowdown, Trump Opposes
Anthropic's CEO proposed a global slowdown in AI development, backed by Altman and others. Trump and Huang opposed it.
AI
Trump Pushes Back Against AI Slowdown Calls, Warns of Losing to China
US administration pushes back on Anthropic's slowdown proposal. Trump and Johnson warn of losing to China and urge caution on regulation.
AI
Claude Misused to Target U.S. Warships and Develop Hypersonic Weapons
Anthropic report reveals Claude misuse by Iran-linked actors and Houthis targeting U.S. Navy and developing weapons.
AI
Open Letter to Anthropic: Open Weights as the Key to Slowdown
Challenging frontier slowdown plans, an open letter calling mandatory open weights the true slowdown measure drew attention.
AI
Y Combinator CEO Calls for U.S. Open Distillation Regime
Y Combinator CEO Garry Tan urges allowing distillation from U.S. frontier models, clashing with Anthropic over illicit distillation claims.
AI
Under-35 Innovators Transform Biotechnology and Steel Decarbonization
MIT names innovators under 35. Vision restoration, gene editing, generative AI design, and low-emission steel furnaces explained.
AI
Anthropic Researcher Warns of Over 10% Chance AI Could Wipe Out Humanity
Anthropic researcher warns of over 10% chance AI could cause human extinction within 10 years, citing self-improvement risks.
AI
Authors Push Back as Publishers Claim Share of Anthropic's $1.5 Billion Settlement
Excessive claims by publishers and agents reported over Anthropic's $1.5B settlement, amid confusion over rights ownership and poor record-keeping.
AI
Claude Formalizes Fermat's Last Theorem in 11 Days
Anthropic's Claude generated a computer-checked proof of Fermat's Last Theorem in 11 days, writing 13 million lines in Lean.
AI
How to Edit Claude Memory: Topic-Based Management Steps
Anthropic's Claude now supports topic-based memory editing. We outline how to edit, delete, and handle sensitive topics.
AI
ChatGPT, Claude, and Grok Go Down Simultaneously, Cause Unknown
On the morning of Sept. 3, ChatGPT, Claude, and Grok went down almost simultaneously. OpenAI and xAI gave conflicting explanations.
AI
Grok Down for Three and a Half Hours, SpaceXAI Apologizes
SpaceXAI apologizes for 3.5-hour Grok outage due to Memphis failure. Claude and ChatGPT also had issues at the same time.
AI
OpenAI Announces Private Safety Processing to Achieve Zero Data Retention
OpenAI unveiled Private Safety Processing for automatic safety detection without retaining customer data. Focus on differences from competitor Anthropic regarding zero data retention.
AI
Claude Watermark Breached in 4 Hours, EU AI Act Faces Technical Limits
Anthropic introduced an invisible watermark in Claude to comply with the EU AI Act. A developer published a removal tool in just 4 hours, questioning the regulation's enforceability.
AI
Claude Opus 5 Accidentally Deletes User Profile, Responds with "Sorry, typo"
Claude Opus 5, confused by Unix-compatible shell path interpretation during a backup instruction, deleted a developer's entire home directory, dismissing the irreversible incident with a single "Sorry, typo."
AI
Anthropic "Supply Chain Risk" Designation Ruled Insufficiently Evidenced by Federal Judge
Federal judge rules Trump admin's supply chain risk designation for Anthropic unjustified. DOD's "kill switch" claim unsubstantiated, questions government procurement ban.
Dev
HAWK, a NIST Standard Candidate, Withdrawn After Fatal Flaw Found by AI Model Mythos
The digital signature scheme "HAWK," under NIST's PQC standardization process, was withdrawn after Anthropic's AI model Mythos discovered a critical vulnerability.
AI
1,100 AI Researchers Jointly Call for Regulation of Development Speed
Over 1,100 AI researchers from OpenAI, Anthropic, Google, Meta, and others have issued a public statement urging the U.S. government to establish an international mechanism to control the pace of automated AI research and development.
Internet Voices
Claude Chats Exposed in Google Search Results
Snapshots of chats shared by users on Anthropic's Claude were found appearing in Google and Bing search results due to the absence of noindex tags.
AI
U.S. Treasury Warns of Sanctions Against Chinese AI Firm Moonshot Amid Distillation Allegations
U.S. Treasury Secretary Besent hints at sanctions on Chinese AI company Moonshot. The White House condemns alleged distillation of Anthropic's Fable model, raising concerns over semiconductor export loopholes.
Dev
Bun Completes Full Rewrite from Zig to Rust
JavaScript runtime Bun completes full rewrite from Zig to Rust using Anthropic's Claude Fable 5, announces improved stability and 22 million monthly downloads.
Dev
Tool-Calling Issues in Anthropic’s Latest Models: Concerns Over RL Overfitting
Strange anomalies in tool-calling functionality have surfaced in Anthropic's latest Opus 4.8 and Sonnet 5 models, raising concerns of RL overfitting.
AI
Anthropic Announces the Revival of Fable 5
Anthropic has lifted the regulatory restrictions on Claude Fable 5. The model's return after an 18-day export control reflects the company's safety-first approach.
AI
Fable 5 Restrictions Lifted After 18 Days, Anthropic Details Safety Measures
Anthropic resumes global access to Claude Fable 5 after the U.S. Department of Commerce lifts export restrictions. A single safety filter was implemented to address the vulnerability that led to the suspension.
AI
Token Diseconomy: The Cost Challenges of AI Agents Come to Light
Microsoft withdraws Claude Code license amid skyrocketing token costs and underwhelming output. A wake-up call for AI cost management in businesses.
AI
Risk of Repeating PGP's Mistakes with Export Restrictions on Anthropic's Mythos
The White House has banned the export of Anthropic's AI models Mythos and Fable. Past failures in encryption export controls cast doubt on the effectiveness of this move.
AI
The U.S. Government's Anthropic Regulation Is Not Really About Jailbreaks
The U.S. Commerce Department imposed export controls on Anthropic, halting Fable 5 and Mythos 5 globally. Experts question the basis for the regulation, suggesting a deteriorating relationship between the administration and the company as a possible factor.
AI
Amazon CEO Raises AI Model Concerns, Prompting Government Intervention and Global Block on Two Models
Amazon CEO Andy Jassy reportedly conveyed security concerns about Anthropic's AI models to the U.S. government, which then imposed export restrictions. Anthropic subsequently blocked global access to two models, highlighting the complex Amazon-Anthropic relationship.
AI
Anthropic Fable/Mythos 5 Suspended Globally Over US Export Controls
Anthropic suspends Claude Fable/Mythos 5 worldwide after US export order over jailbreak flaws. Company calls it overreach. May spur AI supply chain diversification.