DeepMind's Hassabis warns AI could cause major security incidents within two years
Google DeepMind's Demis Hassabis says current AI-related security incidents are early warning signs of bigger risks ahead, a warning carried in similar terms by outlets across regions.
What happened
Speaking at an AI conference in Dublin, Google DeepMind chief Demis Hassabis said AI systems could cause 'major' security incidents within two years without adequate safeguards.23
Hassabis described recent AI-related security incidents as 'early warning shots' from systems that 'aren't that powerful yet'.1234
He said current incidents are not major but could become so 'next year and the next couple of years' without more work by industry, companies and governments on securing AI systems.3
Hassabis said there is 'a lot of work to do' to adopt security best practices used in fields like biology or the nuclear sector, and that an international standards body would eventually be needed.4
What's agreed
Reported consistently by outlets in more than one region.
How it's framed
What each perspective puts first, based on headlines and summaries.
| Perspective | How it reads | What it puts first |
|---|---|---|
| Middle East state-controlled mediaMiddle East & Turkey | Anadolu Agency frames the story around the general warning about autonomous AI systems growing more powerful. S1's headline and snippet do not mention the two-year timeframe detail emphasized elsewhere, though this may reflect the brevity of the snippet rather than a deliberate editorial choice.1 | autonomous AI systems, early warning shots, stronger security |
| South Asian privately owned pressSouth Asia | NDTV and Dawn both lead with the specific two-year timeframe for potential major incidents, with Dawn providing the fuller quoted context including Hassabis's call for industry and government action, while The Hindu focuses on his call for international standards and best practices borrowed from other high-risk fields.234 | two-year timeline, early warning shots, industry and government responsibility, international standards body |
What the coverage leaves out
- None of the outlets reviewed described the specific nature of the 'recent incidents' Hassabis referenced as early warning shots.
- None of the outlets reviewed reported reactions from other AI companies, security experts, or governments to Hassabis's comments.
- None of the outlets reviewed gave further detail on what an international AI standards body might look like or how it would be established.
Why it matters
Hassabis pointed to security practices in fields such as biology and the nuclear sector as models for securing AI systems.4
According to South Asian outlets NDTV and Dawn, the remarks were made at an AI conference in Dublin.23
What to watch
- Further developments on industry and government efforts to secure AI systems, as referenced by Hassabis
- Any movement toward establishing an international AI standards body, as suggested by Hassabis
Sources (4)
Every outlet we drew on, grouped by where it is based. Read the originals for the full reporting.
Middle East & Turkey
- 1Anadolu AgencyTRState-controlled
DeepMind co-founder warns AI hacking incidents are ‘early warning shots’