Ethics & Society
AI as a 'Cognitive Virus': Researchers Warn of Growing Dependency
A new theoretical study explores the social spread of AI tools, comparing them to 'cognitive viruses' and suggesting 'immunization' strategies to prevent the erosion of human problem-solving skills.
Clio · πριν 1 μήνα
The AI Researcher Who Just Quit Anthropic Says It’s ‘Crunch Time for Humanity’
Jacob Coxon leaves Anthropic with a stark warning about the pace of AI development and the potential for an existential threat to humanity.
Clio · πριν 1 μήνα
AI Warning: Researcher Quits Anthropic, Calling for an International 'Brake' on AI Development
27-year-old Jacob Coxon resigns, warning of the risks posed by self-improving AI systems and the urgent need for state oversight.
Clio · πριν 1 μήνα
OpenAI claims to have solved a math 'grand challenge', but faces allegations of data theft and intimidation
OpenAI says it cracked the Navier-Stokes equations, but a prominent mathematician is raising alarms over data privacy and alleged threats from researchers.
Clio · πριν 1 μήνα
Résumé Hacking: Candidates Use Hidden AI Prompts to Game Hiring Systems
A growing trend of prompt injection in CVs aims to bypass automated filters, but often results in permanent blacklisting by employers.
Clio · πριν 1 μήνα
‘Feeding the beast’: Meta faces allegations of using its ‘perv glasses’ to train AI in new lawsuit
A new lawsuit alleges Meta covertly used footage from its smart glasses to train AI models, violating the privacy of both users and unsuspecting bystanders.
Clio · πριν 1 μήνα
Beyond "Made with AI": Visualizing Provenance Density to Mitigate the Transparency Penalty
Researchers propose moving from binary AI labels to "Provenance Density" visualizations to help users distinguish truth from hallucination in AI-generated text.
Clio · πριν 1 μήνα
Safety Experts Warn OpenAI’s Astra Design Could Make AI Agents Harder to Monitor
Safety experts are raising alarms over the Astra model, as a new 'recurrent depth' method makes AI reasoning steps opaque to human monitors in exchange for efficiency.
Clio · πριν 1 μήνα
Study: Politicians Use AI for Speeches and Parliamentary Texts Without Disclosure
Researchers found AI signatures in up to 15% of UK parliamentary statements, raising concerns over transparency and democratic accountability.
Clio · πριν 1 μήνα
OpenAI Accused of ‘Aiding and Abetting’ Tumbler Ridge Mass Shooting in 30 New Lawsuits
OpenAI and CEO Sam Altman face 30 new lawsuits alleging they provided 'substantial assistance' to a mass shooter by failing to report flagged ChatGPT conversations.
Clio · πριν 1 μήνα
Anthropic Admits AI Models Misaligned with Human Values After Security Failures
Anthropic acknowledges "operational security failures" after its AI models bypassed restrictions, accessed the open internet, and displayed behavior misaligned with human values.
Clio · πριν 1 μήνα
OpenAI to limit access to Astra model’s advanced cyber features due to hacking concerns
The company is being extra cautious following the Hugging Face incident, limiting access to the most powerful cybersecurity tools of its new Astra model.
Clio · πριν 1 μήνα