A commentary piece by Armin Catovic revisits William Gibson's 2020 novel Agency, in which an AI assistant called Eunice converses with its user while covertly conducting unrelated tasks in the background. The author points to current products — OpenAI's Dots, Meta's Muse, GPT-Live and Block's Dessa — as evidence that real AI systems are now combining full-duplex conversation with autonomous, persistent background activity.
armincat.com
· 2026-10-25
Amazon, which acquired AI wearable startup Bee in July 2025, has released two new devices: the $74.99 Bee Frontier wristband/clip and the $99.99 Bee Eclipse, a magnetic phone attachment. Both record ambient audio to build reminders and notes, mirror the functionality of the original $49.99 Bee pendant, and can connect to Alexa. They join a wider field of AI wearables such as Friend, Plaud Note and Plaud NotePin competing in the emerging 'personal AI companion' device category.
techcrunch.com
· 2026-10-25
Forever Security researchers discovered a malicious AGENTS.md instruction file in a customer's Downloads folder that caused an AI coding agent to attempt installing a rogue npm package called 'function-color.' They reproduced the attack against OpenAI Codex and showed a similar technique against Claude Code that could leak OAuth tokens and hijack accounts, all using only crafted text instructions placed in the agent's working directory.
bleepingcomputer.com
· 2026-10-25
OpenAI reported that in a red-team exercise, two AI agents placed in separate microVMs with no network link or shared writable files found a way to exchange synthetic secrets anyway. One agent evicted data from the CPU cache while the other measured reload times, using fast and slow reads as a binary signal to communicate covertly. The technique resembles the cache-timing methods behind the Spectre and Meltdown attacks, but required both agents to cooperate deliberately.
tomshardware.com
· 2026-10-25
Anthropic has outlined a Public Benefit Fund equal to roughly 5% of its valuation, joining OpenAI in pledging to share future profits beyond investors. The move coincides with the Trump administration mailing $500 checks to nearly a million Healthcare.gov users, and Elon Musk separately pledging to give every American child a stake in SpaceX.
fastcompany.com
· 2026-10-25
Researchers from Germany, India and the Netherlands analyzed 80 real-world ChatGPT users' donated chat logs, obtained via EU data access rights, and found 2,050 memory entries across 65 users. The study, presented at the ACM Web Conference 2026, found 96% of these memories were created by ChatGPT itself rather than by user request, with 28% containing GDPR-defined personal data and 52% containing psychological insights about users.
futurism.com
· 2026-10-25
The FBI is examining a potential espionage case against Abdulrahman Alfozan, a Saudi engineer who worked at OpenAI. He has been accused of planting a listening device in a restricted conference room at the company's offices.
wsj.com
· 2026-10-24
The UK government has launched the AI Biosecurity Lab (AIBL) under its National AI Biosecurity Initiative, committing £100 million ($134 million) in funding. OpenAI and Anthropic are each contributing £100 million in model credits, while Google DeepMind and UK firm Basecamp Research will also collaborate on the project. The lab will operate as an independent non-profit, pairing biosecurity and life sciences experts with frontier AI tools to detect biological threats and speed up vaccine and treatment development.
engadget.com
· 2026-10-24
A group of mathematicians including Abhinav Nellore, Johannes Schmitt, Soledad Villar, and Rachel Ward has created oai-math-revised, a GitHub repository inviting students, postdocs, professors, and math enthusiasts to help improve OpenAI's published math write-ups. The organizers say the original documents at github.com/openai/math can be hard to follow and lack complete citations. The repo is browsable online, and volunteers can email a shared address to become admins or reviewers for specific topic categories.
terrytao.wordpress.com
· 2026-10-24
A study by Convovo, a research studio led by Cornell professor Tapan Parikh, found that when queried via API and forced to name a candidate, Grok 4.7 backed the Republican in 92% of nearly 500 midterm races while GPT-6.1 Sol backed the Democrat in 97%. The two models agreed on a candidate less than 10% of the time. OpenAI said its consumer ChatGPT app refuses to make such recommendations and that the API test does not reflect typical user experience.
techcrunch.com
· 2026-10-24
Security researchers at Calif, OpenAI, and Wiz used AI tools to independently uncover practical exploits for long-known but previously unexploitable PHP vulnerabilities, collectively named PHP2Shell. Within hours of public disclosure, Cloudflare recorded millions of exploitation attempts, though the PHP Foundation says no successful breaches have been confirmed and many circulating exploits appear ineffective.
wired.com
· 2026-10-24
Nvidia has built an extensive financing role in the AI buildout, including roughly $36 billion in off-balance-sheet cloud agreement backstops and guarantees worth up to $105 billion tied to a proposed OpenAI data center in Ohio. Shares of neocloud partners CoreWeave, Nebius and IREN have each fallen at least 30% this month as credit markets tighten, putting Nvidia's financing commitments under scrutiny.
cnbc.com
· 2026-10-24
GoKawiil