Topic

Anthropic

13 stories

OpinionThe Argument
Anthropic’s self-replication warning needs tests, not automatic acceptance
Anthropic’s multi-node test used emulated Android devices and an unauthenticated port; its separate hardware test showed inference, not autonomous spread.
15 hr ago
Compute
Anthropic says small AI models may soon replicate on connected devices
Anthropic’s eight-node exercise used emulators and an unauthenticated access path; its projected doubling time is measured in days, not observed in real networks.
17 hr ago
PolicyAnalysis
Anthropic’s cruelty ban leaves the line between frustration and enforcement unresolved
The policy takes effect Nov. 12 and relies chiefly on Claude ending conversations, but Anthropic has not published user-impact data or detailed criteria for applying the rule.
Oct 8

More on Anthropic

OpinionThe Human Factor
Anthropic restricts sustained, needless abuse toward its models starting Nov. 12
Anthropic says Claude will keep ending rare, persistently abusive conversations, but the announcement gives no user-impact data or rule-specific appeal process.
Oct 8
OpinionThe Editorial Board
Haiku 5.5’s lower listed rates need independent task tests
Haiku 5.5 starts at $0.10 per million input tokens for prompts up to 100,000 tokens, but published rates and company estimates do not show cost or quality on a buyer’s workload.
Oct 7
ModelsAnalysis
Bijan Bowen says Claude Haiku 5.5 has pretty insane capability; tests used 1% of weekly usage
Bowen said the testing used 1% of weekly usage; the first skate game beat GPT-6 Luna’s, and its revision came close to state of the art from about six months earlier.
Oct 7
ModelsExplainer
Claude Haiku 5.5: what to know about its features, prices and performance claims
Haiku’s token rates drop 90% for prompts up to 100,000 tokens, but Anthropic estimates average running costs fall about 75% after accounting for a tokenizer that uses more tokens.
Oct 7
Compute
Anthropic’s reported EMEA hire signals staffing, not a confirmed data-center build
DatacenterDynamics reported that John M. Wilson will lead EMEA capacity delivery. The report names no site or timetable, and Anthropic’s separate deals do not confirm regional construction.
Oct 7
OpinionThe Argument
Countries outside the U.S.-China AI race need a voice in its rules
Amodei’s case for a democratic lead recognizes authoritarian risks, but U.S.-centered controls need meaningful input from countries that rely on systems they do not control.
Oct 5

Compute
How a voluntary U.S. AI accord enters the global oversight debate
Alex Lo’s opinion column questions who would hold influence in global AI oversight; the U.S. accord’s proposed audits have not been shown to have started or proved effective.
Oct 5
PolicyExplainer
FTC probe of Anthropic and OpenAI: what to know about METR
The agency has confirmed an inquiry into possible consumer risks, but reports of requests to AI evaluator METR and formal evidence demands describe plans, not completed steps.
Oct 2
OpinionThe Argument
Who holds AI firms to account under Trump's self-policing accord?
The one-page pledge Trump and six tech leaders signed promises internal controls and outside audits, but reports go to company boards and no penalty for breaking it has been described.
Oct 1
Policy
Trump's accord leaves 6 AI and chip giants to police themselves, with no penalties spelled out
The one-page pledge commits firms to internal controls, outside audits and board reviews. Trump calls it morally binding, but no penalties, deadlines or public reporting have been described.
Oct 1