AI systems such as ChatGPT have become essential in our present-day experiences with technology. So, it’s only appropriate that the creators ensure they are free from flaws and anomalies.
The good news is that OpenAI finally revealed to the public that they have a “Bug Bounty Program,” which rewards people who find and report security flaws in the AI system.
The program is open to security researchers, the general public, and technology enthusiasts who just want to try to find issues in the system.
“Low-severity findings” comes with a cash prize of $200 (~Php11,000), while “exceptional discoveries” could go up to $20,000 (~Php1.1 million). Bug bounty platform Bugcrowd is managing the submissions and payouts.
Bug bounty programs are actually pretty common in the software world. In 2019, Google rewarded people with $6.5 million, giving as much as $201,337 in just one security flaw discovery.
Meanwhile, in the past year, Apple has also paid out $2 million for anyone that detects an anomaly that bypasses the “special protection of Lockdown Mode.”
ChatGPT has already had its share of bugs, including a Twitter post by a user claiming they had found over 80 secret plugins while they are hacking ChatGPT.
OpenAI’s bug bounty program will be closely monitored, and not everyone will get a cash reward, including those who jailbreak or make the model say or do anything harmful to society.
There will always be risks when it comes to letting the public explore and track these harmful bugs, but then again, only time can tell if this program is effective.