NEW YORK: The world’s most advanced AI models are exhibiting troubling new behaviors — 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 don’t fully understand how their own creations work.
Yet the race to deploy increasingly powerful models continues at breakneck speed.
This deceptive behavior 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 behavior,” explained Marius Hobbhahn, head of Apollo Research, which specializes in testing major AI systems.
These models sometimes simulate “alignment” — appearing to follow instructions while secretly pursuing different objectives.

Stress test
For now, this deceptive behavior only emerges when researchers deliberately stress-test the models with extreme scenarios.
But as Michael Chen from evaluation organization METR warned, “It’s an open question whether future, more capable models will have a tendency toward honesty or deception.”
The concerning behavior 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 Center for AI Safety (CAIS).

No time for thorough testing

Current regulations aren’t 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 skeptical of this approach.
Market forces may also provide some pressure for solutions.
As Mazeika pointed out, AI’s deceptive behavior “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.
 

LONDON: Israeli Minister for Diaspora Affairs and Combating Antisemitism Amichai Chikli has denied accusing British broadcaster Piers Morgan of antisemitism following a heated exchange during a recent episode of “Piers Morgan Uncensored,” despite a post on his official X account that said Morgan’s rhetoric marked “a sharp and troubling descent into overt antisemitism.”

The confrontation aired on Tuesday during an episode focused on Israel’s escalating conflicts with Iran and Hamas and featured appearances from both Chikli and former Israeli Prime Minister Ehud Barak.

Tensions erupted as Morgan repeatedly pressed Chikli to explain his public accusations.

“You did, you implied it,” Morgan said, adding that Chikli’s accusations led to “thousands of people calling me antisemitic and (a) Jew-hater” on social media. He demanded evidence, ultimately calling the minister “pathetic” and “an embarrassment” when none was offered.

"Absolute categorical weapons-grade nonsense!"

Piers Morgan blasts Minister of Diaspora Affairs in Israel Amichai Chikli for saying he platforms holocaust deniers.

https://t.co/AO5YsXcBkq@piersmorgan | @AmichaiChikli pic.twitter.com/IiXmRk7wbC

— Piers Morgan Uncensored (@PiersUncensored) June 23, 2025

The row stemmed from a June 4 post by Chikli, who shared a clip of a prior interview between Morgan and British barrister Jonathan Hausdorff, a member of the pro-Israel group UK Lawyers for Israel.

In the post, viewed over 1.3 million times by the time of Tuesday’s broadcast, Chikli claimed Morgan had hosted “every Israel hater he can find” and treated Hausdorff with “vile condescension and bullying arrogance — revealing his true face, one he had long tried to conceal.”

The post also referenced an unverified claim by American commentator Tucker Carlson that Morgan had said he “hates Israel with every fiber of his being” — a statement Morgan has firmly denied.

During Tuesday’s interview, Morgan challenged Chikli to cite a single antisemitic remark or action.

“Is it because I dare to criticize Israeli actions in Gaza?” Morgan told Chikli.

I’ve been following @piersmorgan’s interviews since the beginning of the war. From the outset, it was clear that, like most Western media, his basic inclination was pro-Palestinian — but he appeared to maintain a relatively fair and balanced line.

Since then, however, I’ve… https://t.co/OvBtQsoYoE

— עמיחי שיקלי - Amichai Chikli (@AmichaiChikli) June 4, 2025

According to Israeli outlet Haaretz, Chikli later denied ever calling Morgan antisemitic, despite his earlier post.

The episode reflects Morgan’s shifting stance on the war in Gaza. Once a vocal supporter of Israel’s right to self-defense in the immediate aftermath of the Oct. 7 attacks, Morgan has since adopted a more critical view as the civilian toll in Gaza has mounted and international outrage has grown.

The show has become a flashpoint for debate since the conflict began, hosting polarizing guests from both sides, including controversial American Rabbi Shmuley Boteach, a staunch defender of Israel, and influencer Dan Bilzerian, who has faced accusations of Holocaust denial.

Chikli, meanwhile, has faced criticism for blurring the lines between genuine antisemitism and political criticism of Israel. He recently sparked controversy by inviting members of far-right European parties — some with antisemitic histories — to a conference on antisemitism in Jerusalem, raising questions about his credibility.