We, and others, have long said that while AI is potentially transformative in a beneficial way, it is also potentially dangerous. Very dangerous.
Europol has teamed up with UNICRI and Trend Micro to examine the immediate threats of AI. The report, available here, does not address the possible existential threat to humanity but rather focuses on the immediate threats to businesses, (their finances and their reputations), governments and consumers.
Criminals, now backed with greater enthusiasm and money by organised gangsters and governments, have always been early adopters of technology.
As Martin Roesler, head of forward-looking threat research at Trend Micro says, “AI is no different. As this report reveals, it is already being used for password guessing, CAPTCHA-breaking and voice cloning, and there are many more malicious innovations in the works.”
The best-known use of AI in the disruption of our lives are deep fakes used as an attack vector but the report identifies this technique as an attack surface as well. New screening technologies will be needed to counter disinformation campaigns and extortion.
Some of the disruption that AI could be used for include:
- convincing social engineering attacks at scale;
- document-scraping malware to make attacks more efficient;
- evasion of image recognition and voice biometrics;
- ransomware attacks, through intelligent targeting and evasion;
- data pollution, by identifying blind spots in detection rules.
The report is not all doom and gloom, however. The core message is collaboration to counter the threats. And use AI as a tool.
In fact, many of the recommendations mirror the recommendations for any cybersecurity issue – multi-disciplinary task forces, de-escalation of political rhetoric and use AI itself to help develop and optimise defences.
Like many transformative technologies before it, AI has huge potential for good. It is already proving beneficial in many industries and many areas of commerce. It has, after all, played an important role in enabling scientists to develop a COVID vaccine at lightning speed.
But, like many technologies before it, the potential for harm is equally huge and equally dangerous.
This report on the real risks and the real remedies against AI-supported attacks is required reading.