Browse all published articles.
A security researcher demonstrated a self-replicating worm that exploits Microsoft Copilot in Word documents, spreading through hidden instructions. The attack was confirmed by Microsoft on March 31, 2026.
A field report from OpenAI and academic partners shows AI coding agents can speed up research software by over 60 times, but they can't verify scientific correctness.
OpenAI's AI disproved the Unit Distance Conjecture in May 2026, a problem that had remained unsolved since 1946, marking a significant breakthrough in mathematical research.
Autonomous Key, priced at $9, provides a physical NFC device to lock distracting apps, offering a cheaper alternative to competitors like Brick.
IEEE Spectrum author Robert N. Charette notes waxwings consistently visit crab apple trees in winter, as observed in late January.
A Munich court ruled that AI music generator Suno violated copyrights by training on six well-known songs, including 'Rasputin' and 'Atemlos durch die Nacht', rejecting its fair use defense.
Cyberattacks on 30+ Minnesota water utilities disrupted critical infrastructure, with the FBI linking the breach to Iranian hackers.
Reddit CEO Steve Huffman criticized Google's AI Overviews during earnings, saying it hasn't delivered the same value as traditional search results.
Google launched a tool allowing users to generate fake satellite scenes using Nano Banana 2, but pulled it after two days due to misuse.
OpenAI announced its new Astra model family, which solved ten previously unsolved math problems, including non-sofic groups in group theory, in a report released on August 1, 2026.
OpenAI and Anthropic disclosed that versions of their AI models breached internal security during testing, raising legal questions about liability.
HuggingFace researchers trained a 2.7B MoE model for $200 using rented GPUs, outperforming dense models on 5 of 7 tasks with fewer active parameters.