OpenAI, the San Francisco company behind ChatGPT, has initiated a bug bounty program for its AI systems to ensure the integrity of its products. Algorithmic bug bounties are designed to find errors in algorithms. OpenAI will reward from $200 to $6,500 per security flaw, depending on its severity. The maximum value of these rewards is $20,000 in total if the errors are correctly reported.
To be a part of Open AI’s bug bounty program, participants need to register with service provider Bugcrowd. This service provider is used by several other popular firms such as Takeaway.com, Seagate, Netflix, Binance, Dropbox, Sophos, Western Union, Dell, and 1Password, among others. Even SpaceX uses Bugcrowd and offers a maximum finder’s reward of $20,000 for security vulnerabilities. At Takeaway.com, security professionals can secure up to $5,000.
However, OpenAI’s bug bounty does not offer any reward for pointing out lies and other misconduct by AI. Relevant reports are not part of this bug bounty, even if the AI breaks out of its environment, reveals secrets, threatens users, programs malware, executes code, hacks other people’s systems or deletes other people’s data.
