Fetching from the wire…
01
02
03
04
05
06
07
08
Research2026-06-27 · source-backed
A June 25 paper (arXiv:2606.25899) argues manipulation capability varies sharply by task rather than being one measurable scalar. (arXiv) That complicates any safety eval trying to score persuasion as a global number. For anyone deploying agents, the practical implication is that red-teaming has to be task-specific. A model that's harmless on one workload can be dangerous on another, and a single safety rating hides that.
Each link below shares sources, entities, or timing with this story.
Executives at Uber, Meta, Microsoft, Salesforce, and DoorDash have launched AI cost-cutting campaigns after bills doubled or tripled, or blew through annual budgets in as little as three to four months. Uber has introduced hard usage limits on AI tools (WSJ). Read that timelin...
arXiv 2609.09553 shows cipher-based covert-communication jailbreaks no longer need fine-tuning on an encrypted corpus. In-context learning is enough, and alignment is significantly weakened or bypassed once the exchange runs through the learned encoding. Demonstrated against m...
Bloomberg reported this morning that Microsoft has begun swapping OpenAI and Anthropic models for its own MAI models inside Excel and Outlook, with tens of thousands of prompts a week now running on MAI. Source. Read that number carefully. Tens of thousands of prompts a week i...
OpenAI's Frontier platform enables building, deploying, and managing AI agents that run other software (Salesforce, Workday, etc.). "Business Context" gives agents institutional memory. Multiyear deals with Accenture, BCG, Capgemini, McKinsey. Key insight: Frontier positions a...
VentureBeat reports OpenAI's GPT-5.6 Sol/Terra/Luna launched to only about 20 vetted partners "per US Gov," the same week Anthropic disabled Fable 5 and Mythos 5. The pattern hardening here is that the most capable models ship behind government-approved allowlists before any p...
Last Week in AI reports Commerce lifted the export-control directive that forced Anthropic to disable both Fable 5 and Mythos 5 for all customers. Anthropic couldn't segregate users by nationality in real time, so a foreign-national restriction became a total shutdown. Engadge...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.