By Key Points
- AI security guidelines will be issued by Singapore this month
- CSA is looking to provide pragmatic ways of securing AI tools
- The guidance must combat fresh and traditional cyber threats
- Industry must work together to improve the security of AI systems
Singapore is launching technical standards that aim to make artificial intelligence (AI) systems more secure. These voluntary rules, which were devised by the Cyber Security Agency (CSA), will be open to public consultation later this month.
Revealed by Janil Puthucheary, senior minister of state for the Ministry of Communications and Information in Singapore, these suggestions are meant to provide practical measures for securing AI tools alongside existing protocols used by companies.
Singapore is working on technical guidelines for securing AI systems https://t.co/NE3a5jz3Xl
— Mrs. Y. (@MrsYisWhy) July 5, 2024
AI’s Growing Threat Landscape
According to Puthucheary, there has been an increase in risk due to the rapid adoption of AI.
He cited McAfee’s tampering with Mobileye’s speed limit signs as one example where adversarial machine learning could disrupt the functionality of an AI system.
The CSA’s guidelines should help reduce such risks by giving cybersecurity professionals a strong foundation on which they can build up the security around AIs.
“Threat landscape has changed significantly with Artificial Intelligence,” said Dr Janil during his speech at Association of Information Security Professionals (AiSP) summit on Wednesday afternoon (May 26).
He added that all parts should collaborate so that we can protect our evolving smart systems from any harm intended against them.
Collaboration and Continuous Improvement
Simulating AI Attacks
The Government Technology Agency (GovTech) is working on being able to simulate attacks against AI systems – which could help understand what kind. and how. and where one may strike these autonomous machines at their weakest point and then put up necessary protection mechanisms around them.
This was shared by Dr. Janil who also pointed out that given their vulnerability towards traditional cyber threats like data breaches; there should always be continuous improvement when it comes to securing these new technologies.
“By simulating attacks, we can better prepare and implement effective safeguards,” he said. This is important as AI continues advancing into different sectors thus increasing its potential attack surface.
Leveraging AI for Cyber Defense
However, AI can also greatly contribute to enhancing cybersecurity where Dr. Janil underlined that artificial intelligence-powered security tools can detect anomalies faster than any other technology so far while still being able to mitigate them accordingly.
He said that machine learning algorithms can identify risks faster and with greater precision, providing security professionals with advanced tools to protect against cyber threats.
AiSP’s Role and Global Efforts
The AiSP is setting up an AI special interest group (SIG) aimed at promoting knowledge sharing among members. Since its establishment in 2008, AiSP has been instrumental in promoting technical excellence within Singapore’s cyber-security community.
The creation of this new group will allow practitioners to exchange ideas on the latest developments in AI and further enhance their collective capabilities.
Singapore’s move reflects global efforts around protecting AI systems from vulnerabilities or compromise.
In April this year, the US National Security Agency’s (NSA) AI Security Center released guidelines for best practices in deploying secure operating systems based on artificial intelligence technology which were done jointly with CISA-USA Cybersecurity & Information Sharing Agency .
These are meant not only to improve reliability but also to deal with known weak points often exploited by attackers when using such platforms.
Singapore’s proactive measures together with its collaborative strategy mark a significant milestone towards safeguarding artificial intelligence technologies.
As more industries continue adopting these transformative innovations, ensuring their safety remains paramount given the ever-evolving threat landscape that they face.
What's your reaction?
