The British intelligence agency GCHQ wants to use artificial intelligence to combat issues such as child sexual abuse, disinformation and human trafficking.
The UK’s adversaries were already using the technology, it said.
The agency has published a paper titled “Ethics of AI: Pioneering a New National Security,” which says it will put the technology at the heart of its operations.
Officially, it says it will help analysts identify patterns hidden in large – and rapidly growing – amounts of data.
This could include:
spotting fake news online that is used by other states to spread disinformation
mapping international networks involved in human or drug trafficking
finding child molesters who hide their identities online.
However, it cannot predict human behavior, such as the execution of a terrorist attack.
The paper also lays out how GCHQ plans to support the AI sector, including setting up an industry-focused AI lab in its Manchester office dedicated to prototyping and mentoring startups.
And GCHQ says it details how it will ensure it uses AI fairly and transparently, including:
an ethical AI code of practice
the recruitment of diverse talent to help develop and guide the use of AI.
It’s a sign that the agency wants to avoid a repeat of the criticism that followed whistleblower Edward Snowden’s revelations that people didn’t know how they were using data.
“While this unprecedented technological development comes with great opportunities, it also presents significant ethical challenges for society as a whole, including GCHQ,” said the agency’s director, Jeremy Fleming.
“Today we set out our plan and commitment to the ethical use of AI in our mission,” he added.
“I hope this will inspire further reflection at home and abroad on how we can ensure fairness, transparency and accountability to underpin the use of AI.”
The report comes as the government prepares to publish its integrated review of security, defense, development and foreign policy, in which technology, including AI, is expected to play a central role.
GCHQ outlined areas where adversaries are already using AI, including:
Foreign state disinformation.
A growing number of states are using AI to automate the production of false content to influence public debate, including “deepfake” videos and audio, GCHQ warns.
The technology can target and personalize this content individually or disseminate it via chatbots or by interfering with social media algorithms.
But it could also help GCHQ detect and review such content and identify “troll farms” and botnet accounts.
Child sexual abuse
GCHQ says AI could:
help analyze evidence of grooming in chat rooms
track the disguised identities of perpetrators across multiple accounts
track hidden individuals and illegal services on the dark web
help police officers infiltrate perpetrator rings
filter content to prevent analysts from being unnecessarily exposed to troubling imagery
As AI is increasingly used to automate cyberattacks, it could also help identify malicious software and attackers who are constantly developing new tactics to penetrate systems and steal data, according to GCHQ.
Organized crime groups are using increasingly sophisticated technologies, including encryption tools, the dark web and cryptocurrencies, according to GCHQ.
But AI could help:
Map the international networks that enable human trafficking – Identify people, accounts and transactions.
“following the money” – analyzing complex transactions and potentially uncovering state sponsors or links to terrorist groups
Merge different types of data – such as imagery and intelligence – to track and predict where illicit goods are being shipped to