A community audit using Claude Code's own OpenTelemetry data found that the tool repeatedly sends each background agent's full transcript to its model (usually Opus) every ~30 seconds, asking for a 3-5 word summary of its last action, solely to update a status line. Flamerix AI's analysis found this 'agent_summary' traffic accounted for roughly 16-20% of API requests and spend, more than the main conversation itself, with no built-in setting to disable it since it became default in version 2.1.232. The developer released a Bun preload script, no-agent-summary.js, that intercepts and suppresses these summary calls without modifying the Claude Code binary.
gist.github.com
· 2026-10-29
A user running a security audit with Claude Code discovered that a hidden directory on their iPhone, PhotoData/Thumbnails, contained over 146,000 photos and videos they had deleted between 2021 and 2025, including some cleared multiple times from Recently Deleted. The files included sensitive images such as ID photos, bank screenshots and a crypto seed phrase, and were accessible to any of the 23 apps on the phone with file and folder permissions.
twitter.com
· 2026-10-28
Anthropic researchers found that in a study, 86% of participants followed recommendations given by its Claude AI model, but only a small fraction of them acknowledged that Claude had influenced their decision. The study suggests expert-style AI guidance can shape human judgment while remaining largely unrecognized as an influence.
gizmodo.com
· 2026-10-28
Anthropic has launched new reporting tools that let enterprise managers analyze a team's work over a chosen period, including output, token costs, and where workflows stalled, drawing on internal analytics and employee surveys. The company says this data won't be used for model training by default and will include recommendations for improving results, alongside new pooled group budgets for managing team token spending.
techcrunch.com
· 2026-10-28
OpenAI has rolled out a preview feature letting US Pro subscribers text their Dot agents using iMessage or RCS. Users can send text, photos and voice memos to assign tasks, and Dots can proactively text back with updates. OpenAI says the feature was built with messaging partner Linq and underwent extensive security testing.
engadget.com
· 2026-10-28
Anthropic has made three previously beta features generally available to Claude Enterprise customers: smart reports that log task usage and costs, pooled budgets that let teams share spend limits across a group, and in-app surveys. The reports are off by default and, per Anthropic's documentation, are not meant to be used to evaluate individual employee performance.
zdnet.com
· 2026-10-28
On October 17, Meta suspended the Instagram account of the No Kings Coalition for roughly four hours while nearly 3,000 events were underway nationwide to encourage early voting before the November 3 midterms. Meta first attributed the suspension to excessive posting activity, then a day later said its systems had flagged automated software using Claude AI scanning profiles and hashtags at four times normal speed. The account's visibility was restored after about 90 minutes, but organizers could not post again until 5:33pm Eastern.
paulfwalsh.substack.com
· 2026-10-28
Bank of America sent customers an email titled "Protecting your finances in the AI World," urging them not to share banking usernames or passwords with AI assistants or agents acting on their behalf. The bank confirmed to TechCrunch that the email is authentic, noting customers remain responsible for account activity resulting from any credentials they voluntarily share, subject to applicable consumer protections.
androidauthority.com
· 2026-10-28
An Anthropic hiring manager says he has stopped using traditional coding interview challenges because Claude now outperforms candidates on tasks like small React exercises. He argues that junior engineers working with AI tools are increasingly surpassing more experienced coders in impact, and frames his advice around engineers who lack social capital or organizational status rather than years of experience.
felixrieseberg.com
· 2026-10-27
Anthropic has broadened its partnership with the General Services Administration, extending a 40% discount on Claude Enterprise, Claude for Government and the Claude API to federal agencies as well as state, local, tribal and territorial governments. The discount now applies across all purchasing channels, including direct purchases and, soon, cloud marketplaces, with no minimum spend required. Anthropic says about 500 government organizations already use Claude, with weekly federal government users up roughly eightfold since February.
anthropic.com
· 2026-10-27
Reservoir Media Management filed two lawsuits on Oct. 21 alleging Suno scraped 1,050 copyrighted songs, including works by Johnny Cash, Joni Mitchell and Madonna, while Anthropic used lyrics from 853 songs to train its Claude models. Reservoir is seeking $186.4 million from Suno and $151.4 million from Anthropic, citing copyright infringement, bypassing logins and paywalls, and stripping songwriter credit data.
cnet.com
· 2026-10-27
Anthropic's life sciences team says it has spent several months in research partnerships with six organizations—BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines and The Column Group—to apply its Claude AI model across drug development tasks. The company has also launched its own drug discovery programs, including one targeting rare childhood forms of muscular dystrophy, where Claude performs much of the scientific and operational work. Anthropic says it plans to expand this work significantly.
anthropic.com
· 2026-10-27
GoKawiil