"AI threats are real - it's not some kind of urban myth": Microsoft UK security head on why your business should be prepared for a new generation of AI cyber threats
THE SO WHAT
When a major vendor’s security lead says AI threats are everyone’s job, that’s a push to treat model abuse and AI-augmented attacks as baseline, not edge cases. This week, make sure your security team—not just your AI team—owns a concrete threat model for prompt injection, data exfiltration via models, and AI-powered phishing.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIInvesco brings its model portfolios into Anthropic’s Claude for advisers
We’re seeing the first real verticalization of frontier assistants into regulated workflows—Invesco, CAIS, Conquest, and FINNY all wiring directly into Claude for Financial Advisors. If you run a product in a regulated domain, assume your distribution will increasingly run through a few assistant surfaces and design your data, permissions, and compliance model accordingly.
What smart people are saying about OpenAI's Exponential Time Hypothesis claim
If ETH really fell, some cryptographic and complexity assumptions would eventually need rethinking—but the immediate story is how quickly labs are now asserting and contesting deep theoretical claims in public. For operators, the move is a reminder to separate marketing around “broken limits” from what’s actually production-ready in your stack.
Amazon tells software engineers what really matters as AI ushers in the era of the 'Renaissance Developer'
Amazon is explicitly telling engineers to ignore AI usage vanity metrics and optimize for business outcomes—expect more large orgs to follow with similar guidance. If you’re leading an engineering team, reframe your AI OKRs around shipped value and defect rates, not prompts, tokens, or “assistant adoption.”
Applied AIThis Japanese brand solves the 2-in-1 laptop dilemma with a 989g frame — offering 28.5 hours of battery life and a stylus charging inside the body
Dynabook’s 989g, 28.5-hour, Core Ultra-based 2‑in‑1 is another data point that “AI PC” differentiation will hinge on battery, thermals, and local models—not just stickers. If you’re speccing fleets for field or hybrid workers, start testing these long-life, on-device-AI form factors against your actual offline workflows.