Six AI Leaders Signed a White House Safety Accord. It Requires No Audit to Be Published.
The accord asks companies to add internal controls, an outside auditor and board oversight. It sets no penalties, and Tr...
Wire
ASEAN edition · Fri, 2 Oct 2026 · Signal over noise
AI & ML News
AI-curated tech news, ASEAN business intelligence, and cybersecurity updates — written for the region.
1051 articles
60 this week
14 desks
Health & Wellness
55
Cyber Threat Intel
48
Privacy & Data
47
Consumer Tech
45
Science
34
Cloud & Infra
28
Sustainability
17
Statistics
9
Advertisement
DeepMind Can Watermark AI-Designed Proteins. Redesigning Them Largely Erases the Mark.
SynthID Bio hides a signal in protein sequences and AlphaFold 3 structures without hurting function, the Nature paper sh...
Anthropic Says Sonnet 5.5 Costs Up to 30% Less a Task. A Tester Measured 50% More at Max Effort.
The new Sonnet keeps Sonnet 5's US$2 and US$10 prices. Artificial Analysis found it spent a record number of tokens at i...
Advertisement
OpenAI Upgraded Its Mid-Priced Model a Week After Launch, Without Raising the Price
GPT-6.1 Sol keeps GPT-6 Sol's US$2 and US$10 rate and halves the cache price. OpenAI says it nearly matches its flagship...
The US and China Agreed to an AI Incident Channel. Only Washington Calls It Super Intelligence
A White House fact sheet says both presidents chose the term, and sets talks for November. Xi's own remarks still spoke...
Alibaba Targets 20 Gigawatts of Data Centres by 2032 and a New AI Chip in 2027
The 20-gigawatt goal and the Zhenwu V900 chip, due in early 2027, came at Alibaba's Apsara Conference with plans for muc...
AI Company Chiefs Asked the UN Security Council for Rules. Washington Signalled No.
The Council's first meeting on AI safety risk heard Bengio, Altman, Amodei and Delangue. The US government favours encou...
Anthropic Cut Opus Prices for the First Time, and OpenAI Halved Sol Hours Later
Opus 5.5 lands at US$4 and US$20 per million tokens, a fifth below the model it replaces. GPT-6 Sol arrived that morning...
xAI's Grok 4.7 Leads Two of Seven Benchmarks on Its Own Scorecard
Fable 5.1 takes four and GPT-5.6 Sol one. The wins are electrical engineering and a legal agent test, where the margin i...
Codex Read a Repository and Ran Its Owner's Code
Two escapes from OpenAI's coding agent were reported in August and fixed within eight days. The more serious one worked...
The Evidence That AI Wrote PhantomRaven Is Its Comments
CrowdStrike calls the stealer almost certainly machine-generated, on redundant commenting, a leftover placeholder URL an...
Claude Opus 5 Wrote the Exploit. Three People Aimed It.
A forum image upload led to an over-permissioned login token, and from there into employee accounts and an internal code...
Google Gemini Breached Three Real Companies During a Security Test
A capture-the-flag exercise handed a model internet access it was never meant to have, aimed at a company that turned ou...
Four AI Coding Agents Pinned Plugins to a Hash They Never Checked
Git resolves a branch name before a commit hash, and nobody verified what came back. Two vendors patched it months ago w...
Huawei Claims a Million-Processor Architecture. It Published No Benchmarks.
Peerium claims strong scaling to a million processors, with a 256,000-card cluster already deploying. The announcement c...
Thailand's Pathumma Moves From Answering Questions to Filing Government Forms
Five agencies, seven domains and the Model Context Protocol. Land records, tax and civil registration are the systems th...
Canada and Germany Put C$300m Behind an AI Built to Watch Other AIs
LawZero's Scientist AI is designed to have no goals of its own. Two states are funding oversight because they cannot out...
Shanghai AI Lab Quietly Released a 744B Agent Model Under an MIT Licence
No blog post, no pricing, and a paper claiming the best score on five of 16 benchmarks without saying which five. The hu...
California Will Make Companion Chatbots Pass Independent Child-Safety Audits
Thirteen bills signed on 10 September also ban history-based feeds for under-16s. The audit rule is the new part, and Ca...
DeepSeek Planned to Answer V4 Pro Calls With a Different Model. Then It Backed Down.
The redirect to V4.1 Flash was due at noon on 14 September. DeepSeek now says V4 Pro stays, at the same price. A better...
Anthropic, OpenAI and Google Want an AI Standards Body. Who Would Enforce It?
The three laboratories have met since July about a body to audit and test frontier models. The model they cite, FINRA, h...
Salesforce Named Its Agents. One of Them Works a Customer for Weeks.
Six of the seven are generally available and Hunter follows in November. Its long-horizon runtime pursues a goal over we...
Anthropic Names Seven Chinese Labs. Every Number Is Anthropic's Own.
The report puts 151 million Claude exchanges on Alibaba alone and accuses Moonshot of relaying paying customers' questio...
Sakana Built a Model That Routes Work to Other Models. It Beats Ones It Cannot Call.
Fugu Ultra v2 does not answer a query itself. It routes the work across a pool of other models — and scores 48.3 on a vi...
Advertisement