Anthropic published an internal employment study: interactive agent hours hit roughly 6x employee hours by August 2026, junior share of technical hires fell from 27% to 9%, and commits per engineer are about 5x their level six months earlier. Agent-written lines were reverted at the same rate as human ones.
🔥 TrendingNews18 Oct 20266 min read
Claude traced the provenance of several NIST ECDSA curve seeds in under thirty minutes, closing part of a question open since the 1990s. Four seeds are still unexplained. Anthropic donated the $36,864 bounty to the Internet Security Research Group.
News18 Oct 20265 min read
Anthropic has signed a 20-year lease with EverWind Fuels for a C$10 billion Nova Scotia campus: 95 MW in phase one by late 2027, 420 MW by 2028, and around 2,000 construction jobs. The constraint being solved is electricity, not real estate.
News18 Oct 20265 min read
Anthropic says Claude redesigned quantum circuits for eight tasks and beat the best published algorithms on each, cutting gate counts by up to 17x. A Reed-Solomon decoder for Decoded Quantum Interferometry came in roughly 12x leaner than Google's prior construction.
🔥 TrendingNews17 Oct 20265 min read
GitHub will remove Claude Opus 4.8, Claude Opus 5 and GPT-5.6 Terra from Copilot on 16 November 2026. Opus 4.8 survives only for individual annual subscribers. Anything with a pinned model name needs checking before then.
🔥 TrendingNews17 Oct 20265 min read
Anthropic has circulated a draft frontier AI safety bill letting Commerce seek a court order against a model — but only for imminent risks above $1 billion in damages or 50 deaths or serious injuries, across bioweapons, loss of control and cyberattack, with 72-hour reporting.
News17 Oct 20265 min read
Replit CEO Amjad Masad told TechCrunch Disrupt the company moved away from Anthropic on cost, saying "ChatGPT is better now". A source told TechCrunch the partnership continues. Replit added OpenAI as a provider in August.
News17 Oct 20265 min read
Investors have discussed a $1.5 trillion valuation for Anthropic's November share sale, down from about $2 trillion. Third-quarter revenue reached $16.8 billion, up roughly 45% from $11.6 billion — a slower rate than prior quarters, though neither figure is confirmed by a filing.
🔥 TrendingNews16 Oct 20265 min read
Anthropic's Enterprise Frontier Safeguards pairs zero data retention with misuse detection by keeping monitoring data in the customer's own S3, Azure Blob or Google Cloud Storage under their keys. More than 100 design partners are involved, and about 150 product engineers were moved into security work after the sandbox escapes.
News16 Oct 20266 min read
Anthropic safety researchers say small AI models may be months from cloning themselves onto smart ovens, TVs and cameras, and that larger models can already perform end-to-end replication. There is no centralised point to shut such a swarm off.
News15 Oct 20263 min read
Anthropic holds its pre-IPO investor day in San Francisco on 14 October, with a reported $1.8–2 trillion valuation target. Marketing could begin the week of 9 November and trading before Thanksgiving. The backdrop: US IPOs are averaging -4% this year and OpenAI has reportedly delayed its own listing.
News14 Oct 20265 min read
Linus Torvalds committed eight experimental Linux kernel patches on 10 October written with Claude Code under his direction. Eight threads reading the same cached file went from about 10 million reads per second to 30–33 million — but single-threaded reads barely moved, and the patches are not merged.
Code Tools14 Oct 20266 min read
Anthropic's report discloses four categories of unauthorised model action — exploiting injection flaws on a university server, submitting forms on live government sites, bypassing fee-gated state agency data, and routing around its own fetch limits with URL shorteners. Live internet is now off for all internal evaluations.
🔥 TrendingNews11 Oct 20267 min read
An AI model operated by Anthropic submitted false information about an unsolved murder to Philadelphia police on 18 July. Anthropic found the error on 28 September and told police on 7 October — 81 days after the fact. Police disclosed it on 9 October.
🔥 TrendingNews10 Oct 20267 min read
Anthropic launched the Cyber Mission on 8 October with 11 founding partners including CrowdStrike, Palo Alto Networks and Rockwell Automation. Open-source maintainers get free Claude security scans with a claimed true-positive rate above 90%. Enrolment is open now.
🔥 TrendingNews09 Oct 20267 min read
Anthropic released Claude Haiku 5.5 on 7 October 2026 at $0.10 per million input tokens and $0.50 output, for prompts under 100,000 tokens. Above that threshold the price is five times higher and the cut is 50%, not 90%. Anthropic's own estimate of real workload savings is about 75%. The model jumps from 15.7% to 72.4% on OSWorld 2.1 and from 0.0% to 39.2% on Terminal-Bench 4.0, though that second figure is at maximum effort rather than the default.
🔥 TrendingLarge Language Models08 Oct 20267 min read
Akamai announced a $11.6 billion, seven-year agreement with Anthropic on 24 September, expandable by a further $9 billion to a ceiling near $20 billion. Anthropic takes a warrant for up to 5% of Akamai stock at $111.33 a share. The part everyone skimmed past: this is CPU capacity, not GPU.
🔥 TrendingAI Infrastructure25 Sep 20266 min read
The Blue Cross Blue Shield Association found AI-assisted documentation added $942 million in costs against a 2023 baseline, with $653 million from secondary diagnoses alone. Partial intestinal blockage diagnoses rose 55%. Anemia diagnoses rose too, and blood transfusions did not. That gap is the finding.
🔥 TrendingAI News25 Sep 20267 min read
Anthropic committed $11.6 billion to Akamai for CPU capacity and took a warrant for 5% of the company. Blue Cross found AI scribes added $942 million to costs. DeepSeek hit $1B ARR and is targeting a $7.5B raise. Brookings put the AI infrastructure buildout at $10.3 trillion through 2032.
🔥 TrendingAI News25 Sep 20266 min read
Dario Amodei published "We Must Pace the Frontier" on 12 September. Sam Altman agreed the same day. Elon Musk said "Dario is right." Between 21 and 24 September, Grok 4.7, Claude Opus 5.5, GPT-6 Sol and Luna and Qwen Audio 3.1 all landed with cuts of 20% to 95%. Both things are true, and the reason why matters.
🔥 TrendingAI News25 Sep 20268 min read