Anthropic's research shows that only 250 malicious files can backdoor large language models, pointing to a critical security flaw that defies assumptions about model size and safety.
Anthropic's research shows that only 250 malicious files can backdoor large language models, pointing to a critical security flaw that defies assumptions about model size and safety.
A critical ASCII smuggling vulnerability in Google’s Gemini AI allows hidden malicious commands in text, but Google has decided against patching it, raising corporate security concerns.
Elon Musk's Grok AI faces scrutiny after antisemitic remarks from the xAI chatbot and a controversial sale to the U.S. government.
Anthropic and OpenAI fund a $450 million startup to test AI models for malicious behavior.
Senator Elizabeth Warren raises concerns over the Pentagon's contract for Elon Musk's Grok AI amid antisemitism issues.
Google Gemini AI is being utilized to enhance tourism in Mexico and combat phishing attacks through new partnerships.
New report details how China’s civilian and military AI integration threatens U.S. strategic dominance.
Anthropic has stopped AI service sales to Chinese-owned firms for security reasons, aligning with broader geopolitical concerns.
New AI technologies are being developed to enhance national cyber defense capabilities.