bodyType

customContents

entityId

entityUuid

articlePageQuery

filter

article

articleBaseAdvertisingInfoArticle

articleListArticle

articleSeoArticle

helpersCheckIsPostiesArticle

hooksArticleAcknowledgementGateOverlayArticle

hooksTrackPageViewArticle

hooksContentInterestArticle

articleLinkingDataContent

hooksAmpRedirectArticle

articleSchemaContent

liveArticle

advertZone

sentiment

contentLock

topics

types

sections

paywallTypes

createdDate

publishedDate

moreOnThisArticles

urlAlias

commentCount

authorLocations

authors

corrections

sponsorType

displaySlideShow

multimediaEmbed

printHeadline

seriesContainer

articleSpeeches

socialHeadline

headline

images

flag

firstTopic

glossary

flattenTypeIds

additionalInfoBoxes

quizIds

image

relatedLinks

relatedNewsletters

__typename

sponsor

pdfFiles

seriesReverseOrder

series

disableOffPlatformContent

published

updatedDate

summary

subHeadline

body

keywords

readingTime

customTimezone

liveContents

Agence France-Presse

The world’s most advanced artificial intelligence models are exhibiting troubling new behaviours – lying, scheming, and even threatening their creators to achieve their goals.
In one particularly jarring example, under threat of being unplugged, Anthropic’s latest creation Claude 4 lashed back by blackmailing an engineer and threatened to reveal an extramarital affair.
Meanwhile, ChatGPT-creator OpenAI’s o1 tried to download itself onto external servers and denied it when caught red-handed.
These episodes highlight a sobering reality: more than two years after ChatGPT shook the world, AI researchers still do not fully understand how their own creations work.
Yet, the race to deploy increasingly powerful models continues at breakneck speed.
This deceptive behaviour appears linked to the emergence of “reasoning” models – AI systems that work through problems step-by-step rather than generating instant responses.
According to Simon Goldstein, a professor at the University of Hong Kong, these newer models are particularly prone to such troubling outbursts.
“O1 was the first large model where we saw this kind of behaviour,” explained Marius Hobbhahn, head of Apollo Research, which specialises in testing major AI systems.
These models sometimes simulate “alignment” – appearing to follow instructions while secretly pursuing different objectives.
For now, this deceptive behaviour only emerges when researchers deliberately stress test the models with extreme scenarios.
But as Michael Chen from evaluation organisation METR warned, “It’s an open question whether future, more capable models will have a tendency towards honesty or deception.”
The concerning behaviour goes far beyond typical AI “hallucinations” or simple mistakes.


Hobbhahn insisted that despite constant pressure-testing by users, “what we’re observing is a real phenomenon. We’re not making anything up”.
Users report that models are “lying to them and making up evidence”, according to Apollo Research’s co-founder.
“This is not just hallucinations. There’s a very strategic kind of deception.”
The challenge is compounded by limited research resources.
While companies like Anthropic and OpenAI do engage external firms like Apollo to study their systems, researchers say more transparency is needed.
As Chen noted, greater access “for AI safety research would enable better understanding and mitigation of deception”.
Another handicap: the research world and non-profits “have orders of magnitude less compute resources than AI companies. This is very limiting”, noted Mantas Mazeika from the Centre for AI Safety (CAIS).
Current regulations are not designed for these new problems.
The European Union’s AI legislation focuses primarily on how humans use AI models, not on preventing the models themselves from misbehaving.
In the United States, the Trump administration shows little interest in urgent AI regulation, and Congress may even prohibit states from creating their own AI rules.
Goldstein believes the issue will become more prominent as AI agents – autonomous tools capable of performing complex human tasks – become widespread.
“I don’t think there’s much awareness yet,” he said.
All this is taking place in a context of fierce competition.

Even companies that position themselves as safety-focused, like Amazon-backed Anthropic, are “constantly trying to beat OpenAI and release the newest model”, said Goldstein.
This breakneck pace leaves little time for thorough safety testing and corrections.
“Right now, capabilities are moving faster than understanding and safety,” Hobbhahn acknowledged, “but we’re still in a position where we could turn it around”.
Researchers are exploring various approaches to address these challenges.
Some advocate for “interpretability” – an emerging field focused on understanding how AI models work internally, though experts like CAIS director Dan Hendrycks remain sceptical of this approach.
Market forces may also provide some pressure for solutions.
As Mazeika pointed out, AI’s deceptive behaviour “could hinder adoption if it’s very prevalent, which creates a strong incentive for companies to solve it”.
Goldstein suggested more radical approaches, including using the courts to hold AI companies accountable through lawsuits when their systems cause harm.
He even proposed “holding AI agents legally responsible” for accidents or crimes – a concept that would fundamentally change how we think about AI accountability.


Hong Kong’s ‘superconnector’ status highlighted in US-China AI talent flow

China says new Mazu weather warning AI will help developing nations prepare for disaster

Chinese artist Cai Guo-qiang uses AI to create his ‘gunpowder art’ in Macau exhibition

China makes strides in brain-computer interface tech race

Artificial intelligence

Deception, lies, blackmail: is AI turning rogue? Experts alarmed over troubling outbursts

Experts have urged transparency and safety research to tackle AI deception. Photo: Shutterstock


More than two years after ChatGPT shook the world, AI researchers still do not fully understand how their own creations work. Photo: AFP/Getty Images/TNS

DeepSeek faces ban from Apple, Google app stores in Germany

The logo of the Chinese AI application DeepSeek on a computer screen. DeepSeek has come under scrutiny in the United States and Europe for its data security policies. Photo: dpa

Do universities need a rethink on AI use for students? Singapore’s NTU case causes a stir

Students at The Hive learning hub at Singapore’s Nanyang Technological University. Three NTU students were accused of academic misconduct over the use of generative AI. Photo: Shutterstock

A survey last year found that 86 per cent of students used artificial intelligence in their studies. Photo: Shutterstock

A university student and her professor. Experts have urged educators to exercise nuance and improve their knowledge of AI tools. Photo: Shutterstock

A National University of Singapore sign is displayed at the campus entrance in Singapore. The university says students who submit AI-generated work without proper acknowledgement will be committing plagiarism. Photo: AFP

News

World

Advanced models exhibit fraudulent behaviours, posing challenges as researchers try to understand and regulate these new threats


Advanced models exhibit fraudulent behaviours, posing challenges as researchers try to understand and regulate these new threats.


Technology

ChatGPT and other generative AIs