Amazon’s new bug bounty program aims to engage external security researchers in identifying vulnerabilities within its NOVA AI models. The initiative invites select third-party researchers to explore common security threats affecting generative AI systems, providing a financial incentive for discoveries related to prompt injection and model manipulation, including potential misuse in producing weapons.
Participants in this invite-only program will be chosen by Amazon, ensuring stringent control over which researchers gain access to its technology. Historically, Amazon has compensated researchers over $55,000 for validated AI-related vulnerabilities under its broader bug bounty efforts. With generative AI increasingly woven into its services, including Alexa and AWS, the company emphasizes the importance of maintaining robust security protocols to prevent real-world exploitation and improve overall safety in AI technologies. This introduction follows earlier initiatives, including a competition that awarded $250,000 to 10 research teams for examining AI model vulnerabilities, resulting in significant findings around safety and functionality trade-offs.
👉 Pročitaj original: CyberScoop
