OpenAI has announced a partnership with Hacktron, a security firm, to find vulnerabilities in OpenAI's own systems. The work covers three areas: probing OpenAI's infrastructure for weaknesses, testing the connections between its platform and third-party services, and using frontier AI models to strengthen the company's defences. The move reflects a broader push by AI developers to test their own products before attackers do. As large language models are connected to more outside tools, data sources and business applications, each integration adds potential entry points. Third-party testing can surface flaws that internal teams may miss, and using advanced AI in the process suggests OpenAI sees its models as useful for defensive security work as well as a source of new risk. The announcement does not disclose contract terms, the scope of testing, or any vulnerabilities found so far. For Australian businesses and professionals, this changes nothing directly. It does not alter obligations under the Privacy Act or any current contract. It is worth noting for teams that use OpenAI's products or APIs, because it shows the vendor is investing in security testing of its third-party connections. Organisations that rely on these tools should keep applying their own vendor due diligence, review what data is shared with AI services, and check the security documentation and access controls they have configured.