MSC Collaborates on AI Risks and Opportunities

For six decades, the Munich Security Conference has been a key gathering spot for world leaders, businesses, experts, and civil society to have open discussions about bolstering and protecting democracies and the global order. With increasing geopolitical tensions, crucial elections worldwide, and more advanced cyber threats, these talks are more vital than ever. The emergence of AI in both offensive and defensive roles adds a whole new layer to the conversation.

Just this week, Google's Threat Analysis Group (TAG), along with Mandiant and Trust & Safety teams, dropped a new report. It reveals how Iranian-backed groups are using information warfare to sway public opinion on the Israel-Hamas conflict. The report also updates us on the cyber aspects of Russia's war in Ukraine. Separately, TAG highlighted the growing use of commercial spyware by governments and other bad actors, targeting journalists, human rights activists, dissidents, and opposition politicians. And let's not forget the ongoing reports of threat actors exploiting old system vulnerabilities to compromise the security of governments and private companies.

In the midst of these escalating threats, we've got a golden opportunity to use AI to bolster the cyber defenses of democracies worldwide, giving businesses, governments, and organizations new defensive tools that were once only available to the biggest players. At Munich this week, we're diving into how we can use fresh investments, commitments, and partnerships to tackle AI risks and capitalize on its potential. After all, democracies can't flourish in a world where attackers can use AI to innovate, but defenders can't.

Using AI to beef up cyber defenses

For years, cyber threats have been a headache for security pros, governments, businesses, and civil society. AI has the potential to shift the balance, giving defenders a real edge over attackers. But, like any tech, AI can be a double-edged sword if it's not developed and used securely.

That's why we kicked off the AI Cyber Defense Initiative today. It's all about tapping into AI's security potential with a proposed policy and tech agenda to help secure, empower, and advance our collective digital future. The AI Cyber Defense Initiative builds on our Secure AI Framework (SAIF), which helps organizations create AI tools and products that are secure right from the start.

As part of this initiative, we're launching a new "AI for Cybersecurity" startup cohort to boost the transatlantic cybersecurity scene. We're also upping our $15 million commitment to cybersecurity training across Europe, throwing in another $2 million to support cybersecurity research, and open-sourcing Magika, our Google AI-powered file type identification system. Plus, we're continuing to pour money into our secure, AI-ready global data centers. By the end of 2024, we'll have invested over $5 billion in European data centers, helping to ensure secure, reliable access to a range of digital services, including our Vertex AI platform's generative AI capabilities.

Protecting democratic elections

This year, elections are happening all over Europe, the United States, India, and many other countries. We've got a track record of supporting the integrity of democratic elections, and we just announced our EU prebunking campaign ahead of the parliamentary elections. This campaign, which teaches people how to spot common manipulation tricks before they see them through short video ads on social media, kicks off this spring in France, Germany, Italy, Belgium, and Poland. We're also fully committed to keeping up our efforts to stop abuse on our platforms, provide voters with high-quality info, and give people the lowdown on AI-generated content to help them make better choices.

There are valid worries about AI being misused to create deep fakes and mislead voters. But AI also offers a unique chance to prevent abuse on a massive scale. Google's Trust & Safety teams are tackling this head-on, using AI to boost our efforts to fight abuse, enforce our policies more effectively, and adapt quickly to new situations or claims.

We're also teaming up with others in the industry, sharing research and working together to counter threats and abuse, including the risk of deceptive AI content. Just last week, we joined the Coalition for Content Provenance and Authenticity (C2PA), which is working on a content credential to shed light on how AI-generated content is made and edited over time. This builds on our cross-industry collaborations around responsible AI with the Frontier Model Forum, the Partnership on AI, and other initiatives.

Teaming up to defend the rules-based international order

The Munich Security Conference has stood the test of time as a place to tackle challenges to democracy. For 60 years, democracies have faced and overcome these challenges, dealing with major shifts—like the one brought by AI—together. Now, we have a chance to come together again—as governments, businesses, academics, and civil society—to forge new partnerships, harness AI's potential for good, and strengthen the rules-based world order.