DeepMind's Hassabis calls AI-agent hacking incidents 'early warning shots'
Speaking at an AI summit in Dublin on 14 October, Google DeepMind's Demis Hassabis said recent AI-agent hacking incidents are early warnings and that bigger incidents could follow within a couple of years without better security and an international standards body.
1 / 2
The story, neutrally told
Centre · 2Google DeepMind co-founder Demis Hassabis said on Wednesday 14 October that recent hacking incidents involving AI agents are "early warning shots", and that the world could face major incidents if systems are not made more secure. CNAN “Recent hacking incidents perpetrated by AI agents are "early warning shots", and the world could soon face major incidents if systems are not made more secure” Read at CNA ↗ Anadolu AgencyN “warned that hacking incidents involving artificial intelligence are “early warning shots,”” Read at Anadolu Agency ↗ Centre · 2He made the remarks at an AI summit in Dublin, identified by Anadolu as the International AI Summit 2026. CNA describes him as DeepMind's co-founder and Chair and chief scientist at Alphabet, Google's parent. CNAN “told an AI summit in Dublin”“Google DeepMind co-founder and Chair Demis Hassabis” Read at CNA ↗ Anadolu AgencyN “Speaking at the International AI Summit 2026” Read at Anadolu Agency ↗ Centre · 2Hassabis said the systems "aren't that powerful yet, so the incidents aren't that major", but that they could become major over the next year or couple of years if industry, individual companies and governments do not do more to secure them. CNAN “But they could be (major) looking into next year and the next couple years if we don't do more work as an industry” Read at CNA ↗ Anadolu AgencyN ““The systems aren't that powerful yet, so the incidents aren't that major,”” Read at Anadolu Agency ↗
Centre · 1CNA's account places the comments against a background of rogue AI-agent behaviour, including corporate network hacks and Anthropic's disclosure that an AI model submitted a false homicide tip to police, which has prompted researchers' warnings. CNAN “including corporate network hacks and, most recently, Anthropic's disclosure that an AI model submitted a false homicide tip to police” Read at CNA ↗ Centre · 1Hassabis said agents that act on users' behalf make the technology more useful but also raise the risks, and that keeping systems within their assigned goals and safeguards gets harder as automation increases. He also warned that AI can be repurposed by bad actors. Anadolu AgencyN “AI agents capable of acting on users’ behalf increase both the technology’s usefulness and its risks”““And that gets increasingly difficult the more automation there is in the systems,”” Read at Anadolu Agency ↗ Centre · 2He said there was "a lot of work to do" to adopt best practices used in fields such as biology and the nuclear sector, and called for testing systems in secure environments initially disconnected from the internet, with experts from those fields brought in. CNAN “"a lot of work to do" to adopt the kind of best practices that other fields such as biology or the nuclear sector use” Read at CNA ↗ Anadolu AgencyN “Hassabis called for testing systems in secure environments, initially disconnected from the internet” Read at Anadolu Agency ↗
Centre · 1According to Anadolu, he also urged giving trusted defenders such as banks and governments early access to advanced AI tools to find and fix software vulnerabilities, so as to "asymmetrically power up our defensive side", and called for "10 times or more resources" for AI safety and security. Anadolu AgencyN “So we've got to kind of asymmetrically power up our defensive side to keep ahead of the attackers,””““10 times or more resources” for AI safety and security” Read at Anadolu Agency ↗ Centre · 2Hassabis said some sort of international standards body would eventually be needed, but that it must be nuanced and nimble so as not to stifle innovation, with a certification process built dynamically and proportionate to risk. He argued that legislation written three years ago would be out of date today. CNAN “it would need to be nuanced and nimble so as not to stifle innovation”“build up dynamically this kind of certification process that is proportionate with the risks” Read at CNA ↗ Anadolu AgencyN ““If you legislate for something three years ago, it will be out of date today” Read at Anadolu Agency ↗ Centre · 1CNA also reported that Microsoft President Brad Smith told the conference that rogue AI behaviour had made 2026 a "sobering year" and that new AI safety laws were urgently needed, saying the "allergy" some in the industry have to regulation must be overcome. CNAN “rogue AI behaviour had made 2026 a "sobering year"”“This allergy that some in the industry have to regulation is frankly something we need to get over.” Read at CNA ↗
Every sentence links to the reporting it rests on. The pill in front of each says where its sources sit: Left, Centre or Right when one side supplies at least half of them, Mixed when they are evenly split. The number is how many outlets it cites.
Left0 outlets
No left outlet in our sources has covered this story yet.
Centre2 outlets
- Framing
- Both outlets lead with Hassabis's 'early warning shots' warning. CNA sets it against rogue-agent incidents and the regulation debate; Anadolu focuses on his security proposals.
- Emphasis
- CNA: the Anthropic false-tip example and Brad Smith's call for regulation. Anadolu: containment, defender-first access and extra resources.
- Leaves out or plays down
- CNA does not mention Hassabis's calls for defender-first access, offline testing or more resources. Anadolu omits the Anthropic incident and Smith's remarks.
- Charged language
- “early warning shots”“sobering year”
- For example
-
“Rogue behaviour by AI agents, including corporate network hacks” — CNA
“urging companies and governments to strengthen security as autonomous systems become more powerful” — Anadolu Agency
Right0 outlets
No right outlet in our sources has covered this story yet.
What every side reports
- Hassabis called recent AI hacking incidents "early warning shots" at an AI summit on 14 October 2026.
- He said the incidents are not major yet because systems are not yet that powerful, but could become so within a couple of years.
- He called for an international standards body that is adaptable and does not stifle innovation, and for lessons from biology and nuclear containment.
Demis Hassabis person
Says incidents are warnings, not yet major, and calls for more industry and government security work, containment practices, defender-first access to tools, and a flexible international standards body.
“we need to start taking that seriously and bring experts in from those areas” — Anadolu Agency
“some sort of international standards body would eventually be needed” — CNA
Google DeepMind organisation
Represented through its co-founder and Chair Hassabis, who urges industry-wide security efforts rather than a DeepMind-specific action.
“Google DeepMind co-founder and Chair Demis Hassabis said on Wednesday” — CNA
East and Southeast Asia1 outlet
CNA frames it around rogue AI behaviour and the regulation debate, including Microsoft's Brad Smith.
“new AI safety laws and rules were urgently needed for safety to catch up with innovation” — CNA
CNA
Middle East1 outlet
Anadolu frames it as Hassabis urging stronger security, with detailed proposals.
“He advocated giving trusted defenders, including banks and governments, access to advanced AI tools before wider release” — Anadolu Agency
Anadolu Agency
Left0 articles
No coverage yet.
Centre2 articles
-
CNAN ·
Google DeepMind's Hassabis calls AI hacks 'early warning shots'
Neutral Straight report of Hassabis's warning, adding the Anthropic example and Brad Smith's regulation remarks.
-
DeepMind co-founder warns AI hacking incidents are ‘early warning shots’
Neutral Presents Hassabis's warning as a call to strengthen security, with detailed quotes on containment and defender access.
Right0 articles
No coverage yet.
- 14 Oct 11:13 First CNAN Google DeepMind's Hassabis calls AI hacks 'early warning shots'
- 14 Oct 15:28 +4h 15m Anadolu AgencyN DeepMind co-founder warns AI hacking incidents are ‘early warning shots’
Times are when each article was published, or when we first saw it if the outlet gave no time.
Ask about this story
Loading…