Category: AI
Lasso Study Finds SynthID-Text Can Alter LLM Safety Behavior
Lasso researchers report that Google DeepMind’s SynthID-Text watermarking can affect model refusals, prompt injection and agent choices.
AI Job Fears Rise Among Undergraduates in Global Survey
A Global Undergraduate Awards survey found 55% of participating students viewed AI as a concern, up from 48% in the previous year.
Hacktron Used Claude Opus 5 to Reach OpenAI Code in Bug Bounty Test
Hacktron researchers reportedly used Claude Opus 5 and an image-processing flaw to access an OpenAI account and internal GitHub repositories.
AI BioDesign Uses AI and Lab Testing to Explore Molecules Beyond Nature
The Allen Institute-led AI BioDesign project is developing models and experimental methods for creating and testing novel molecules.
AI Risk Debate Divides the American Left Over Regulatory Priorities
A dispute over existential AI risk is splitting US progressives between frontier-model safeguards and action against present-day harms.
Microsoft Memo on AI Scraping Surfaces in New York Times Lawsuit
A Microsoft scientist’s 2023 memo warned about AI scraping as court filings raise questions about publisher traffic and training data.
Anthropic Reports Attempts to Use Its AI Models for Bioweapon Research
Anthropic says users explored dangerous pathogen modifications with its AI, renewing debate over safeguards for biological research tools.
Scientists Challenge Claims That AI Could Independently Build Bioweapons
Scientists say AI can accelerate access to biological knowledge, but physical labs, expertise and materials still constrain bioweapon risks.
AWS, Qualcomm and chipmakers refocus AI infrastructure on inference
AWS, Qualcomm, d-Matrix, Oracle and Broadcom are adapting processors, memory and networking to meet rising enterprise AI inference demand.
Anthropic Publishes Three Metrics for Tracking the Pace of AI Development
Anthropic has detailed methods for measuring AI-led research, autonomous agent oversight and the allocation of computing capacity.
SIGNAL TITANS