OpenAI teams up with Cathedral to find & fix AI flaws faster, boost bug bounties

As artificial intelligence (AI) becomes smarter, the tools used by hackers to attack it are also getting more advanced. To stay ahead of these cyber threats, ChatGPT-maker OpenAI is gearing up for a new internal "security sprint". The company is taking aggressive steps to test its digital walls, and to do this, it has partnered with Cathedral Systems, a cybersecurity startup known for its fast and effective hacking methods.Joining hands with Cathedral SystemsOpenAI has announced that its upcoming internal security sprint will be conducted in partnership with Cathedral Systems. The main goal is to test and strengthen OpenAI's defenses using AI itself. The company sees this as a continuous effort to make its systems hack-proof by using model-assisted testing.Cathedral Systems is a relatively new player in the cybersecurity world, founded by former DOGE staffers. They recently proved how capable they are when their AI agents successfully found security gaps in rival AI firm Anthropic's third-party business systems in under 12 hours.By teaming up with Cathedral Systems, OpenAI hopes to use their unique skills to spot hidden weaknesses in its own network before any bad actors can take advantage of them. Massive reward for finding bugsTo make this major security drive a success, OpenAI is asking outside experts and hackers to help find flaws in its software. The company has significantly increased its bug bounty rewards, which means it will now pay a lot more to people who find critical issues.According to the new update, anyone who makes an exceptional discovery will be eligible for a payout of up to $172,000, including bonuses. This is more than four times the previous reward for critical findings.Learning from the previous security sprint These security sprints are a big deal for the company. During its last sprint, OpenAI dedicated 25 per cent of its product engineers just to finding and fixing security weaknesses. In that exercise, engineers used Codex to look into the findings, reproduce the vulnerabilities in safe test environments, and create software patches to fix them.OpenAI says this is not just a one-time thing, but a continuous, ongoing process to keep its systems safe.What OpenAI shared on XIn a detailed update shared on X (formerly Twitter), OpenAI highlighted the importance of using AI for digital defense. The company noted that since AI is getting better at finding vulnerabilities, the good guys need to use it to secure systems before attackers can exploit them.OpenAI explained that their partnership with Cathedral Systems will allow external researchers to use OpenAI's own models alongside internal teams to challenge their current defenses. This collaboration will bring a fresh approach to finding and fixing weaknesses using AI tools.The company also confirmed the huge increase in its bug bounty program. By more than quadrupling the rewards for critical bugs, they are offering up to $172,000 for truly exceptional findings.As AI models become more powerful, keeping internal secrets and user systems safe is a top priority for tech giants. With this massive bounty hike and expert partnership, OpenAI is making sure its digital defenses are just as smart as the AI it builds.- Ends