US, Britain, other countries ink agreement to make AI ‘secure by design’

In a groundbreaking international agreement, the United States, Britain, and a coalition of 16 other countries have come together to address the pressing issue of keeping artificial intelligence (AI) safe from rogue actors. In a 20-page document, the countries highlighted the need for AI systems that prioritize security from the design phase. While the agreement is non-binding and lacks specific directives, it emphasizes the importance of monitoring AI systems for abuse, safeguarding data from tampering, and thoroughly vetting software suppliers.

Jen Easterly, the director of the U.S. Cybersecurity and Infrastructure Security Agency, expressed her satisfaction with the agreement, stating that it marks a shift away from prioritizing “cool features” and fast market release, towards prioritizing the safety of AI systems. However, it’s worth noting that this agreement is just one of many initiatives by governments worldwide attempting to shape the development of AI, and few of these initiatives have any real teeth.

The agreement addresses concerns about AI technology falling into the wrong hands and being exploited by hackers. It includes recommendations on releasing models only after rigorous security testing. However, it does not delve into complex issues surrounding the ethical uses of AI or the collection of data that fuels these models.

While Europe has taken the lead in AI regulations, the United States has been lagging behind due to a polarized Congress. The Biden administration has called for AI regulation to protect consumers, workers, and national security. However, progress has been slow. The hope is that this international agreement will serve as a stepping stone towards more comprehensive AI regulations in the future.

Key Points:
– The United States and Britain, along with 16 other countries, have unveiled an international agreement on AI safety.
– The agreement emphasizes the importance of prioritizing security in AI systems from the design phase.
– The agreement is non-binding and lacks specific directives, but it highlights the need to monitor AI systems for abuse, protect data, and vet software suppliers.