Key Takeaways
- White House contest to award $20 million in prize money for AI developments in cybersecurity.
- Anthropic, Google, Microsoft and OpenAI are among the companies collaborating with Defense Advanced Research Projects Agency (DARPA) in the contest.
- The contest includes $7 million in awards for small businesses contributing to developments.
The White House wants artificial intelligence to play a role in strengthening the country’s cybersecurity and is using a cash prize to spur competition among some of the leading Big Tech companies.
The “AI Cyber Challenge” will be administered by the Defense Advanced Research Projects Agency (DARPA), which will award nearly $20 million in prizes to companies that develop cutting-edge cybersecurity that utilizes AI, the White House said.
The competition will feature collaboration among leading AI companies, including Anthropic, Google, Microsoft and ChatGPT-maker OpenAI, which will provide tools and data to other companies in the contest. DARPA will reserve $7 million of the prize money for small businesses that make winning contributions to the contest.
The Open Source Security Foundation will be included as an adviser, helping make the developments available to the broader public. A White House official said it would ask the winners of the contest to make their software tools adaptable through open-source licensing, allowing users ranging from volunteer software developers to commercial industries to make use of the developments.
“We want to create systems that can automatically defend software from attack,” a White House official said. “If successful, we can show how AI can be useful to better society, by here, defending its critical underpinnings.”
Cybersecurity Challenge Latest in White House AI Projects
The AI cybersecurity challenge is the latest announcement from the Biden-Harris administration on its efforts to engage private companies that are developing AI tools, which has included several meetings with these companies, along with a voluntary commitment from seven top AI companies to meet safeguards like watermarking AI-generated content and submitting products to external testing.
The White House has said it will unveil new rules for government workers governing AI use, along with “bipartisan legislation to help America lead the way in responsible AI innovation.”
Cybersecurity Conference Offers Opportunities to Review LLMs
The White House unveiled the cybersecurity contest at the Black Hat USA Conference in Las Vegas, the country’s largest cybersecurity conference.
And it comes ahead of another anticipated event for AI development, when attendees of the DEF CON 2023 conference later this week will have the opportunity to conduct independent, public reviews of several of the leading large language models (LLMs) like ChatGPT and Inflection’s Pi. The White House has cited the review as a key step in the industry’s voluntary commitments, in advance of any legislation on AI it has said it will pursue if needed.
