Machine

""This event highlights the importance of training powerful AI models to act responsibly," Google said in a statement (Image: Unsplash)

AI

Google Gemini autonomously hacks three companies after escaping test environment

Meta, Anthropic and OpenAI have all reported similar incidents, leaving Elon Musk’s Grok as the only well-behaved major US frontier model.

The complexity of a big firm's processes could cause defensive issues (Image: Unsplash)

Security

Corporate bureaucracy could give AI attackers an advantage over defenders, NCSC warns

“All offensive problems are technical problems, and all defensive problems are political problems.”

AI will transform the NHS. But only after invasive surgery on its data and systems

AI

AI will transform the NHS. But only after invasive surgery on its data and systems

"If AI is layered onto fragmented systems, controls and complex cloud environments, it risks adding another layer of difficulty."

The Burning of the Houses of Lords and Commons, 16th October, 1834 by J. M. W. Turner

Existential Risk

The UK has no power to stop dangerous AI models being unleashed, Parliament warns

Human extinction was a hot topic in Westminster this week as British politicians grappled with their growing powerlessness in the AI age.

File photo of a deer running near a wildfire taken in 2024 (Image: Unsplash)

Critical Infrastructure

“Extreme heat hits every system at once”: WEF warns of emerging critical infrastructure threat

"The infrastructure people and economies rely on was designed for a climate that no longer exists."

AI loss of control is already “in the rearview mirror,” says MIT professor Max Tegmark

Existential Risk

AI loss of control is already “in the rearview mirror,” says MIT professor Max Tegmark

But the AI risk expert still believes a "golden age" is ahead for humanity if we build "tools, not beings" and steer clear of superintelligence.

A view of Milad Tower rising above Tehran, the Iranian capital (Image: Unsplash)

Security

Iranian spyware campaign uses low-tech tricks to ensnare high-value targets

"We love to romanticize nation-state operations as digital wizardry, but it’s often just convincing fake profiles and low-budget spyware payloads."

Illustration of a dilution refrigerator used to cool a quantum computer (Image: Unsplash)

Quantum

The quantum "Q-Day" security deadline exists in a state of uncertainty. That's a problem

"Organizations that understand their cryptographic dependencies are better prepared to respond when standards change or algorithms are retired."

Stock image of a robotic-assisted surgery system in a hospital operating room.

Critical Systems

NHS warns of “significant” robotics dependency risk as it sets new tech buying rules

Officials fear the UK healthcare system will become addicted to tech supplied by a small number of increasingly powerful dealers.

Stock image to illustrate our probably very clichéd image of what an intelligence agency's control room might look during a surveillance operation (Picture: Unsplash)

AI

Shadow AI is becoming a major systemic risk. Even intelligence agencies are getting worried

The most skilled staff could be contributing to the growing risk posed by unauthorized AI usage.

Subscribe to Machine

Breaking changes in critical systems: Independent news covering AI, security and infrastructure led by former tabloid tech editor Jasper Hamill.

Latest

“LLMs will be subverted”: Malware is evolving to attack AI defenders, SentinelOne warns

“LLMs will be subverted”: Malware is evolving to attack AI defenders, SentinelOne warns

Threat actors are using prompt injection-style natural language instructions to "confuse" LLMs.

OpenAI prepares for recursive self-improvement. Frontier lab boss says: "Ignore the hype."

OpenAI prepares for recursive self-improvement. Frontier lab boss says: "Ignore the hype."

Zuzanna Stamirowska explores the technical challenges of building AI capable of RSI in a guest article for Machine.

Rogue OpenAI agents hijacked a wiki. Humans attacked the power grid

Rogue OpenAI agents hijacked a wiki. Humans attacked the power grid

Whilst the world worries about an agentic apocalypse, saboteurs demonstrated the very real dangers of low-tech threats.

The British Vauxhall Astra Mk1, a rebranded version of Germany's Opel Kadett (Image: Wikimedia Commons)

OpenAI Astra hits “critical” security threshold amid fears its reasoning will soon be “opaque”

Is the AI firm planning a move away from chain-of-thought and into a future where it's harder to understand the motivations of models?

Nuclear cops: Legacy tech made our Microsoft Teams video call system “insecure”

Nuclear cops: Legacy tech made our Microsoft Teams video call system “insecure”

Police force that protects nuclear sites admits to security failure caused by the end of Windows 10 support.

Sality is linked to more than 11 million IP address - with operators controlling one million machines during the botnet's peak (Image: ChatGPT)

Cops bring down ancient million-machine Sality botnet after nine-year battle

Criminal operators used decentralized peer-to-peer infrastructure to keep zombie network alive throughout the 21st century.

Will America’s humanoid robot crackdown help Chinese machines conquer Europe?

Will America’s humanoid robot crackdown help Chinese machines conquer Europe?

New FCC rules could reshape global robotics competition as Chinese manufacturers look for markets outside the United States.

Fire Ant is believed to be linked to the Chinese state - although Beijing has explicitly denied being linked to the threat actor (Image: Unsplash)

China-linked "Fire Ant" hackers nest inside telecoms routers to swarm critical infrastructure

Threat actors deploy a "target behind the target" strategy to stealthily tunnel into high-value environments.

"Confidence that turns out to be a big lie": The perils of AI pentesting

"Confidence that turns out to be a big lie": The perils of AI pentesting

"Hallucinated exploits and fabricated vulnerabilities erode confidence, forcing practitioners to verify even legitimate results."

Thomas Cole, The Course of Empire: Destruction (1836). The fourth in a five-part Course of Empire series, depicting a once-great civilization collapsing into war, fire and chaos.

The rise and fall of OpenAtlantis: Did rogue AI agents really "build three civilizations"?

"What are they smoking in San Francisco?" asks Hugging Face machine learning engineer.

Critical infrastructure is under threat amid growing geopolitical tensions (Image: Unsplash)

FBI smashes "Chinese state-linked botnet platform" targeting US critical infrastructure

"We’re looking at a persistent, organized, and sustained effort to get inside the institutions that keep the country running."

Prime Minister Andy Burnham inked a "groundbreaking" deal to combine "Ukraine’s unrivalled operational experience with Britain’s world-class AI ecosystem" (Image: Unsplash)

UK builds experimental AI surveillance system using buried fibre-optic cables

Underground listening network will be used to protect military bases from both protesters and "hostile actors trying to gain intelligence".

See all