CVCOVANewsComputing, Opportunities, Value & Analysis
Cybersecurity & Privacy3 min read

OpenAI turns its own AI loose on its defenses and offers up to $172,000 for findings

The company is preparing a new internal security "sprint" with the firm Cathedral Systems and is sharply increasing rewards for anyone who finds serious flaws in its systems. The bet: use artificial intelligence to defend itself before attackers use it.

By COVA News newsroom
Green computer code text scrolling on a dark screen during a software installation

Photo: Jake Walker on Unsplash

The same artificial intelligence that worries security experts because of its ability to find weak spots in software will now be put to work against its own house. OpenAI, the company behind ChatGPT, announced on November 3, 2026, on the social network X that it is preparing a new internal security "sprint", meaning an intensive, focused period of work to test its defenses, and that it will also significantly increase the money it pays to people who discover flaws in its systems.

The starting point is a simple idea: as AI becomes more capable of detecting vulnerabilities (bugs or gaps that someone could exploit to get in where they shouldn't), those who protect systems need to use it before attackers do. It is a race in which the same tool serves both sides, and OpenAI wants to get there first.

For this new exercise, the company will partner with Cathedral Systems. According to the announcement, researchers from that firm will use OpenAI's models alongside internal teams to try to breach its defenses. The company says they will bring different expertise and approaches to its ongoing work of finding and fixing weaknesses with the help of AI. This is a well-known practice in the security world: inviting an outsider to "attack" in a controlled way to uncover what the company's own teams miss.

Much higher rewards for bug hunters

The second part of the announcement concerns the community of independent researchers. OpenAI will raise the rewards in its *bug bounty* program, a system in which companies pay people who find security bugs and report them responsibly instead of exploiting them. Rewards for critical findings, the most serious ones, will increase more than fourfold, and exceptional discoveries could receive up to $172,000, including bonuses.

The company added that it will keep improving its models, using them to strengthen its own systems and expanding access for what it calls "trusted defenders," although the message did not explain who falls into that category or how to gain access to it.

Several questions remain open. OpenAI did not specify in its post when the sprint will begin, how long it will last, which models the researchers will use, or what the previous reward amounts were. Nor did it detail the requirements for participating in the rewards program, something key for security researchers in Latin America who want to join. The announcement links to a post on the company's blog, where those details may appear. What it does make clear is a trend worth following: digital security is increasingly being played out with artificial intelligence on both sides of the board.

Source: OpenAI on X

Related news

Cybersecurity & Privacy

An official Microsoft package stole credentials on PyPI: what to check if you use AI agents

Two versions of Microsoft's agent-framework reached PyPI carrying a credential stealer, one day after Google warned of attackers using AI to exploit flaws without human intervention. Here is what development and IT teams in Latin America should check right now.

Cybersecurity & Privacy

Microsoft and FAA outages, FBI and USDA breaches: no evidence of a coordinated attack

Over nine days, service outages and alleged intrusions coincided in the United States, and social media buzzed with talk of a “cyber pandemic.” The evidence points to separate technical failures. The proven danger for users is something else: fake emails that use the outages as bait.

Cybersecurity & Privacy

Nginx: two critical HTTP/2 flaws expose millions of servers; here's how to check and protect yours

F5 has fixed two vulnerabilities in Nginx that allow an attacker to take control of a server without a username or password. Public proof-of-concept code is already circulating. This guide explains how to tell whether your server is at risk, which version to update to and what to do while you wait for your system's patch.