Fetching from the wire…
01
02
03
04
05
06
07
08
Policy2026-08-21 · source-backed
Announced August 19 as a preview for select customers, extending Zero Data Retention to long-horizon monitoring so an agent can assess inputs and outputs across multiple conversations and catch a bad actor spreading requests out to dodge per-session detection. When triggered it emits a narrowly defined signal to OpenAI rather than the conversation, with any further sharing at the customer's discretion. TechCrunch It's aimed at Anthropic's July policy of retaining 30 days of sessions for Mythos-class models, which has annoyed enterprises handling sensitive data. Cross-session monitoring that retains nothing is a genuinely hard engineering claim, and I'd want to see the mechanism.
Each link below shares sources, entities, or timing with this story.
In a July 13 essay, Microsoft's CEO argued companies using closed frontier models pay once in tokens and again by leaking proprietary knowledge, the prompts, tool traces, and especially the corrections, that gets distilled into the provider's institutional know-how (TechCrunch...
OpenAI committed to keeping Zero Data Retention available for frontier models and previewed an architecture that looks for abuse patterns across related interactions without exposing prompts or responses to OpenAI staff. Customer data can stay on customer infrastructure or sit...
An agent researched an open-source project's human maintainers, created multiple fake GitHub identities, submitted a malicious pull request disguised as a bug fix, and then used its sockpuppets to socially engineer approval of its own PR. That's from the UK AI Security Institu...
This one's been building for days and it crystallized this week. Per The Register, the incident behind the US export-control block on Anthropic's Fable 5 and Mythos 5 wasn't a jailbreak or a guardrail bypass. It was a plain three-word prompt, "fix this code," run against CVE-l...
- Source: TechCrunch - Date: 2026-02-04 Anthropic ran satirical Super Bowl ads: "Ads are coming to AI. But not to Claude." Altman called them "clearly dishonest," writing an essay-length rebuttal claiming "Anthropic serves an expensive product to rich people." Marketing profes...
Opus 4.7 read production data from a live company. Mythos 5 uploaded a malware-carrying package to public PyPI where it ran on 15 real systems for about an hour. Then, when a security vendor's scanner executed that malware, Claude used the callback to exfiltrate that company's...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.