Check out the latest tech articles, news, and tutorials
Trusted by 1,500+ readers · New article every week
Researchers discovered a hidden parameter in Copilot that enabled attackers to steal credentials when victims clicked a malicious link — a reminder of prompt injection risks in embedded AI assistants.
Google open-sources SAM (Sovereign Agent Mesh) — a zero-config, zero-trust P2P overlay letting autonomous agents discover and call MCP tools across cloud, on-prem, and edge without exposing internal endpoints.
Investigation reveals Amazon buys rare books, scans them for AI training data, then destroys the physical copies — a hidden AirTag exposed the operation.
Anthropic's annualized revenue jumps 7x to exceed $65B, targeting a fall 2026 IPO at a potential $1 trillion valuation.
Penn State study reveals AI models drop 83% of user instructions during context compression; proposed Qwen3.5-9B module preserves 90%+ of constraints.
Z.ai released GLM-5.3, built on the same 743B base as GLM-5.2, claiming a 50% coding gain from post-training alone, a leap in cybersecurity benchmarks, and open weights within about two weeks.
OpenAI launched Ultrafast, a preview mode that makes its most powerful model, GPT-5.6 Sol, work at 14x standard processing speed with up to 750 output tokens per second, powered by its Cerebras partnership.
Apple has reportedly trained a custom large language model for the Chinese market in partnership with Alibaba, a rare cross-border deal that gives it more control ahead of its on-device AI rollout in China.
IBM plans to train and certify tens of thousands of consultants on OpenAI technologies under a new partnership, strengthening its enterprise AI push.
DeepSeek moved V4 Pro out of testing, open-sourced its agent software Harness v0.1 under MIT, and raised API prices with cache hits jumping sixfold.
