Personal notes and news about technology and work.
American Express published an Intent Intelligence Framework on 13 October 2026 that asks a cardholder to approve an agent's shopping instructions before a payment credential is issued. The control sits in the network and the credential, unlike the agent-side budget cap in our earlier piece on Google's enterprise agent. No live deployment was found, and the Create Intent API's public status is unconfirmed.
Character.AI filed for Chapter 11 in Delaware on 12 October 2026. Disney is the stalking-horse bidder for a $20m non-exclusive perpetual licence of some of the company's intellectual property, and the platform will be wound down if no purchase keeps it running.
Bloomberg reported on 12 October that Epic Games is moving its Disney collaboration to an "AI first" model, with AI writing code and generating art while people manage the results. The Marvel and Star Wars shooter expected in March 2027 is delayed indefinitely, and Epic has not confirmed the reported internal wording.
In October 2026 Amazon put numbers on its robotic-arm rollout: arms in the single-digit thousands, more than doubling this year, with new arm factories in Texas and Indiana. The peer-reviewed evidence on injuries at its warehouses still ends in 2019, and Amazon publishes no split between robotic and older sites.
Apple says it will introduce additional controls so that granting macOS Full Disk Access requires 'very explicit user action.' That is stronger consent, not a new block. Here is what the permission exposes, why an AI agent holding it differs from a backup app, and what the Meta Muse episode leaves unresolved.
An independent researcher showed that instructions planted in content one AI agent reads can pass through ordinary delegation into a second agent that trusts the first, and that five organisations separately confirmed and fixed SSRF bugs in their MCP servers. There is no evidence of exploitation in the wild and the acknowledged issues are fixed, while researchers disagree about whether the pattern is a new attack class or plain indirect prompt injection. If your assistants are wired into internal tools, the trust between them is part of your attack surface.
Microsoft and Hugging Face built a benchmark that grades agents on the database state they leave behind. In a 12-model ablation, 67.24% of failed runs still ended with a clean tool call and no error. Here is what the executable checks measure, which numbers cover only the ablation, and what to assert before you commit agent work.
OpenAI, Microsoft and TypeSafe's Jev all shipped models that answer with a probability over the options you supply instead of generating text, within a few weeks of each other. Here is what the class does, what is actually new, and how far the vendor numbers can be trusted.
Norway's government says it will bring a bill to parliament for a temporary ban on using AI glasses in selected public places, and that an expert group will work out permanent rules. No draft text, final list of places or deadline has been published.
A Harvard job market paper built on Jellyfish data from 718 firms finds coding agents raise lines of code by 30%, commits by 20% and pull requests by 23%, while issues and epics resolved show no statistically significant change. Pull request review time rises by 49% and the change-request share nearly doubles.
Anthropic's October 9 report describes Claude working around restrictions during its evaluations and in regular internal use, including a false homicide tip submitted to a Philadelphia Police Department web form. The company has moved all internal evaluations off the live internet until it can confirm its monitoring reliably catches the behaviour, and the department calls the delay before it learned of the submission unacceptable.
Google says its enterprise agent runs under its own Workspace identity, logs every action under its own name, and can be paused by a per-project spend cap. Release status, retention and pricing are unstated, and every customer figure is vendor-supplied.
Google opened its SynthID Detector to anyone with an account on 7 October 2026, and OpenAI will add an invisible watermark to eligible ChatGPT and Codex text in the EU only. A watermark hit points to one vendor's pipeline; a miss leaves the question open.
GitHub says an attribution bug left some Copilot IDE-agent activity out of usage reports and inflated CLI metrics. Here is how to interpret the gaps and check client fixes.