AI & RoboticsNews

Why GPT-4 is vulnerable to multimodal prompt injection image attacks

OpenAI’s new GPT-4V release supports image uploads — creating a whole new attack vector making large language models (LLMs) vulnerable to multimodal injection image attacks. Attackers can embed commands, malicious scripts and code in images, and the model will comply.  Multimodal prompt injection image attacks can exfiltrate data, redirect queries, create misinformation and perform more…
Read more
AI & RoboticsNews

MonsterAPI leads the charge in democratizing AI with no-code fine-tuning

A new platform that allows users to fine-tune open-source large language models (LLMs) without writing any code has recently launched new features that make the process even easier and faster. The platform, called MonsterAPI, was created by a team of researchers and developers who wanted to make LLMs more accessible and affordable for everyone. LLMs are powerful artificial intelligence systems…
Read more
DefenseNews

The Army wants to see more ‘creativity’ from special operations forces

Above all else, the commander of 1st Special Forces Command says he wants to see creativity out of the Army’s special operators — he thinks it’s what will provide them with an edge in the next fight in light of the rapid tactical innovations occurring on Ukraine’s battlefield. Brig. Gen. Lawrence ”Gil” Ferguson, who spoke with Army Times, is a firm believer in an oft-repeated special…
Read more
DefenseNews

Navy aims to hasten remaining Columbia sub work in case of test delays

DANVILLE, Va. — The U.S. Navy is working to accelerate remaining construction work on the first Columbia-class ballistic missile submarine to brace for potential delays during the first-in-class boat’s testing later this decade. The Navy and prime contractor General Dynamics’ Electric Boat originally built a six-month buffer into the construction schedule for the future District of Columbia…
Read more