OpenAI will pay you to hunt for ChatGPT bugs

ChatGPT is a wildly well-liked show of the wonders of generative AI. However the utility nonetheless has numerous hurdles to leap, significantly with privateness points and buggy conduct. To assist mitigate these points, OpenAI opened its Bug Bounty Program to crowdsource insightful bug-finding.

In partnership with Bugcrowd, OpenAI is asking for moral hackers to search out vulnerabilities within the software program and report them to the corporate. OpenAI additionally needs these hackers to check ChatGPT plugins to search out authentication points, authorization points, safety points, and an inventory of different points.

Additionally: What’s ChatGPT and why does it matter?

OpenAI’s Bug Bounty Program additionally asks hackers to search out out if delicate OpenAI data could possibly be uncovered to 3rd events, reminiscent of Notion, Asana, Salesforce, and plenty of others. Hackers are requested to solely discover in-scope points, as out-of-scope points aren’t eligible for money prizes.

Out-of-scope questions of safety embody jailbreaking or bypassing the chatbot’s security options, getting the chatbot to faux to code (OpenAI says if ChatGPT executes code, it is partly or totally hallucinating), or tricking the chatbot into telling you dangerous issues.

If an moral hacker is to search out an in-scope problem, they are often eligible to win as much as $20,000. OpenAI does not disclaim precisely which vulnerability discovery is price $20,000, however the jackpot discover does apply to a couple classes.

Additionally: How you can use ChatGPT

If somebody finds a very exceptional vulnerability inside API targets, ChatGPT logins and subscriptions, or vulnerabilities within the OpenAI analysis group’s web site and providers, they may win a $20,000 prize.

Decrease precedence vulnerabilities can win between $200 and $600 money prizes, center precedence vulnerabilities can win a hacker between $600 and $1,250, and excessive precedence vulnerabilities can win somebody between $1,000 and $3,500 relying on which in-scope class the vulnerability falls below.

Hackers with regarding hacking finds can submit them to OpenAI’s Bugcrowd web page. Per OpenAI’s Bugbounty settlement, hackers that discover information leaks are stored confidential till OpenAI authorizes their launch.