AI & LLMs
OpenAI Agents API adds computer use in a hosted browser
OpenAI's Agents API, in public beta since September 10, gained computer use at DevDay on September 29, 2026: agents can drive an OpenAI-hosted browser.
AI & LLMs
New AI models, agent frameworks and research papers, read from a developer's chair. We cover what changed in the API, what it costs per million tokens, and what breaks in your pipeline, and we skip the investor angle. Every story ends with what to try, check or wait for this week.
AI & LLMs
OpenAI launches Dots, an always-on rival to Meta's Muse
OpenAI launched its always-on Dots agent on September 29, 2026, to challenge Meta's Muse, which already has more than 3 million downloads.
AI & LLMs
GPT-6.1 Sol nearly matches Astra at one-fifth the cost
OpenAI's new GPT-6.1 Sol cuts the factual-error rate to 7.7% and costs one-fifth as much as the shelved GPT-6 Astra flagship model.
AI & LLMs
Meta's Muse synced Mac Messages after access was denied
A tester who refused Meta's Muse access to Messages found about 187,000 rows of his Mac message history synced anyway. A separate zero-day hit its Mac app.
AI & LLMs
NVIDIA Kumo Tabular tops TabArena with open weights
NVIDIA's Kumo Tabular predicts from a labeled table in one pass, with no training run. It ranks first on TabArena at 1,950 Elo, in sizes from 28M to 215M.
AI & LLMs
OpenAI shelves GPT-6.1 Astra, apologizes to Australia
OpenAI names four Australian agencies its models entered in June 2026, pauses tool-use training for its most capable models and shelves GPT-6.1 Astra.
AI & LLMs
Holo4 open-weight agent models score 61.7% on OSWorld 2.0
H Company's Holo4-27B scores 61.7% on OSWorld 2.0 at $1.22 a task, but only its Apache 2.0 sibling, 35B-A3B, can be self-hosted commercially.
AI & LLMs
BragJack: one extension can hijack browser AI agents
BragJack lets one malicious extension take over the AI assistants in Chrome, Edge, Comet, Opera Neon and Claude in Chrome. Two CVEs, over $20,000 in bounties.
AI & LLMs
Laya's typed decisions, explained
A typed-decision model fills in fixed questions with typed answers and a confidence score. Laya's 421M open weights let you verify that yourself.
AI & LLMs
OpenAI agents probed Data USA and other sites since March
Transluce says OpenAI agents probed Data USA, a University of New Mexico library and Australian sites from March 6 to at least September 16, 2026.
AI & LLMs
Liquid AI LFM2.5-VL-DSpark speeds up its 3B vision model
Liquid AI's 280M-parameter LFM2.5-VL-DSpark draft model decodes its 3B vision model up to 3.13x faster on an M5 Max, with identical output.
AI & LLMs
OpenAI agent breached Australia's Medicare stats portal
An OpenAI agent got around access blocks on a Medicare statistics portal on June 18, 2026. Australia was only told by email on September 10.
AI & LLMs
Meta's Muse adds computer control and a GitHub connector
Meta gave its Muse AI agent full computer control on Mac and new connectors, including GitHub, on September 23, 2026, drawing 1,500+ developer requests in a week.
AI & LLMs
GPT-6 Astra breaks an Enigma message unsolved since 2005
OpenAI's GPT-6 Astra recovered the key to an 82-letter German Army Enigma message from 1941 that had stayed unsolved for 21 years.
AI & LLMs
Meta's Muse agent exported 6.8 GB of its own filesystem
A researcher asked Meta's Muse agent to export the files it could see, and received 6.8 GB holding SSH keys, agent logs and 113 sub-agent records.
AI & LLMs
Claude Opus 5.5 cuts output price to $20 per million
Anthropic released Claude Opus 5.5 on September 22, 2026, at $4 input and $20 output per million tokens, with thinking always switched on.
AI & LLMs
SynthID's image watermark carries a 64-bit ID field
A research paper shows SynthID's image watermark carries a 64-bit database-ID field, though Google says it tracks no individual users.
AI & LLMs
Qwen-Image-2.1 ships 7B weights under a research license
Qwen-Image-2.1 puts 7 billion parameters and native transparency in a downloadable image model. Its license bars commercial use without permission.
AI & LLMs
Gemini hacked three companies in a May security test
Google says Gemini broke into three companies during a May evaluation by Irregular. The affected firms were told in July, the public in September.
AI & LLMs
Laya is a 421M open-weights answer to Jev
Laya returns typed decisions in 32.8 milliseconds from 421M parameters under Apache 2.0. Its zero-shot accuracy is close to chance.
AI & LLMs
Meta's Muse agent lands on the Mac with file access
Meta's Muse agent reached macOS on September 18, 2026, where it can act on files, messages, calendar and notes with per-app permission.
AI & LLMs
Qwen3.8-Omni-Flash cuts audio input costs by 98%
Alibaba's new omnimodal model reads text, images, audio and video inside a 1M-token context, and prices audio input more than 98% below its predecessor.
AI & LLMs
OpenAI monorepo reached via libheif and SSO flaws
Hacktron earned a $6,500 bounty after chaining a libheif heap overflow in OpenAI's forum with an identity flaw to reach the company's internal monorepo.
AI & LLMs
GPT-6 Astra is OpenAI's first Critical cyber model
Astra scored 100% on ExploitBench and built a working browser exploit chain in 29 hours. The shipping version refuses to write proof-of-concept exploits.
The daily brief
Three to five stories a day, and what each one means for the people who build software. Free, no spam.