TheChek News Flow

The former Disney CEO has dabbled in tech and media investing

Bob Iger is returning to venture capital as he charts his next act after running Disney.

The former chief executive officer of Disney, who stepped down from his role last month after a nearly two-decade run, has taken an advisory role at Thrive Capital, the venture firm founded by Joshua Kushner, according to people familiar with the matter. He will work with the firm’s staff on investments and with founders of companies in Thrive’s portfolio, one of the people said. Thrive, which recently raised $10 billion for a new set of funds, has backed the likes of Instagram, Spotify, A24 and OpenAI.

Kushner is a former Goldman Sachs banker and son of real-estate developer Charles Kushner. He is also the younger brother of Jared Kushner, the son-in-law of President Trump.

Iger’s next move has been the subject of much speculation in Hollywood. The longtime media executive, whose previous attempt at retirement ended abruptly when he returned to the top Disney job, has invested personally in startups and at one point toyed with the idea of running for public office. He is expected to remain on Disney’s board of directors through December.

For Iger, who owns a stake in Thrive, the move is something of a homecoming.

He first stepped down as CEO of Disney in 2020 after a 15-year stint to make way for handpicked successor Bob Chapek. Iger briefly joined Thrive as a venture partner in 2022, but relinquished the role when he returned to lead Disney later that year after Chapek’s firing by the board of directors.

During his time at Thrive, Iger met OpenAI CEO Sam Altman. Last year, Iger helped orchestrate a $1 billion deal with OpenAI, by which Disney agreed to license more than 200 of its characters so users could create AI-generated content with OpenAI’s text-to-video offering Sora. That deal fell apart after OpenAI decided to scrap Sora.

News Corp, owner of The Wall Street Journal, has a content-licensing partnership with OpenAI.

Between his two stints running Disney, Iger was also involved in a number of entertainment and technology startups. He and his wife Willow Bay, dean of University of Southern California’s communication and journalism school, bought a controlling stake in women’s soccer team Angel City FC in 2024.

OpenAI on Thursday released GPT-5.5, its newest AI model, which the company calls its “smartest and most intuitive to use model” yet. The algorithm comes with increased capabilities in a multitude of areas, with OpenAI co-founder and president Greg Brockman claiming that it also brings the company one step closer to the creation of OpenAI’s “super app.”

On a call with journalists, Brockman said that the new model was a big advancement “towards more agentic and intuitive computing.”

“This model is a real step forward towards the kind of computing that we expect in the future — but it is one step, and we expect to see many in the future,” Brockman said. “It’s a faster, sharper thinker for fewer tokens compared to something like 5.4. So this means that there’s just more frontier AI available for businesses and for consumers, which is part of our goal.”

Brockman also said that the model was an additional step toward creating a “super app” — a multi-purpose, Swiss Army knife of a program — which Brockman and co-founder Sam Altman have previously discussed launching. The co-founders envision combining ChatGPT, Codex, and AI browser into one unified service that can aid enterprise customers. Notably, the “super app” concept is also a hot topic with Altman rival (and former OpenAI colleague) Elon Musk, who has said he wants to turn X into its own so-called super app.

OpenAI released its last model only last month, with a previous release in December and, before that, November. The company has continued to churn out new models at a crisp pace, a trend that company staff said should be expected to continue for the foreseeable future. “We see pretty significant improvements in the short term, extremely significant improvements in the medium term,” said Jakub Pachocki, OpenAI’s chief scientist. “In fact, I would say, like, I think the last two years have been surprisingly slow.”

According to OpenAI, GPT-5.5 is designed to be useful across a broad array of categories, including foundational enterprise areas like agentic coding and knowledge work, but also in more experimental AI applications like mathematics and scientific research. The company also released data Thursday showing the model’s superior performance across a range of benchmarks. Compared to its previous models, and to models from competitors Google and Anthropic (like Gemini 3.1 Pro and Claude Opus 4.5), 5.5 consistently scores higher, according to OpenAI.

OpenAI’s rivalry with Anthropic is never far from discussion, and GPT-5.5 offered another opportunity for the two companies to compare themselves to one another. One reporter during the press briefing asked if GPT 5.5. would have capabilities similar to Mythos, the cybersecurity tool recently announced by Anthropic. (Mythos has experienced controversy in recent days due to a report of unauthorized access to the program.) Mia Glaese, a member of OpenAI’s technical staff, said that GPT-5.5 would have a significant impact on the company’s approach to deploying its models toward digital defense. “We have a strong and longstanding strategy for our approach to cyber, and we’ve refined a durable approach to rolling out models safely,” Glaese said.

Mark Chen, chief research officer at OpenAI, said that GPT-5.5 was better at navigating computer work than its predecessors, and also said that the model “shows meaningful gains on scientific and technical research workflows,” noting that the company feels it could really “help expert scientists make progress.” Chen also said it could assist with drug discovery, an area that has shown increased industry interest over the last few years.

GPT 5.5 is widely available starting Thursday, according to OpenAI. The company says that the model is deploying to Plus, Pro, Business, and Enterprise users in ChatGPT, while 5.5 Pro is headed to Pro, Business, and Enterprise users.

Hours after Anthropic said it was investigating a report that users had gained unauthorized access to its ballyhooed Mythos model, OpenAI accidentally made a slate of its own unreleased models available on its Codex app.

The breaches are a reminder that Anthropic and OpenAI have plenty of their own cybersecurity issues even as they help other organizations fortify their systems for a new era of AI-powered cyberattacks.

Anthropic has gone to great lengths in recent weeks to stress that Mythos is capable of devastating cyberattacks, and that it was only making the model available to a select few companies and government agencies as a result.

Even so, a group of unauthorized users had been accessing Mythos without Anthropic’s permission and sharing their findings in a Discord channel, Bloomberg reported Tuesday. Anthropic said in a statement that it was investigating the apparent leak, which it believes was made possible by software that made the model available to a third-party contractor.

The OpenAI leak included models named GPT-5.5, arcanine, glacier-alpha and Heisenberg, the last of which was labeled “latest frontier life science research model,” according to a screen recording a Codex user posted online. A person with direct knowledge said the leak was a significant lapse.

“Last night, for less than 20 minutes, a very small number of Codex users were able to access a limited set of non-public models due to a temporary configuration error,” said a spokesperson for OpenAI in a statement. “We quickly remediated the issue, conducted a full investigation and found no evidence of malicious or concerning use during this period. No internal systems or source code were accessed.”

Security experts say a leak like the Mythos one was all but inevitable. They’re now warning that defenders should assume Mythos—or similarly capable models—will be available to hackers imminently and should start to prepare their cyber defenses accordingly.

“Most of us believed that it was only a matter of time before we had to confront this new reality, whether it was unauthorized access or the models becoming part of the public domain,” said Andrew Rubin, CEO of cybersecurity startup Illumio.

Rubin’s company has been testing OpenAI’s forthcoming cybersecurity model, GPT-5.4-Cyber, which OpenAI also has declined to release publicly to give defenders time to prepare for the types of attacks it could facilitate. He said defenders are still figuring out how to respond to the growing AI security threat, but in the meantime companies should focus on segmenting their IT systems so that if hackers breach one part they can’t easily get into others.

“Whether it’s models from Anthropic, OpenAI, or other firms not located in the U.S., models are going to keep getting much better than the human brain has been at finding vulnerabilities, and that’s not going to slow down,” Rubin said. “I don’t think anybody really has an answer for what the operating model should look like in this new world.”

Part of the urgency from cybersecurity experts is that, even without cutting edge models like Mythos, existing models are already extremely good at sophisticated cyberattacks. A rash of hacks and breaches have hit companies including Mercor, Vercel, and the open source project Axios in recent weeks, which researchers suspect have been fueled by AI models.

“Those of us who pay attention to incident response have noticed an extreme pace of offensive operations in recent months,” said Alex Stamos, a Stanford cybersecurity researcher who previously served as Facebook’s chief security officer and is now an executive at coding security startup Corridor. “It's hard to tell whether this is because of AI or a coincidence, but it’s very possible we’re already seeing the effects of AI on hacking.”

Tech executives have long warned of this. In a 2024 essay, Mark Zuckerberg used the poor state of security in the AI industry to argue in favor of open-source AI models that anyone can download. “Some people argue that we must close our models to prevent China from gaining access to them, but my view is that this will not work and will only disadvantage the U.S. and its allies,” he wrote. “Our adversaries are great at espionage, stealing models that fit on a thumb drive is relatively easy, and most tech companies are far from operating in a way that would make this more difficult.” Those comments look mighty prescient now.

Buck Shlegeris, CEO of Redwood Research, which studies cybersecurity approaches for containing AI models, said it is reasonable to think that hackers have already stolen the list of security vulnerabilities that Anthropic found using Mythos.

“Anthropic is currently not robust to high-effort security threats,” he said on The Information’s TITV this week. “I kind of have the attitude that if you aren’t going to be able to secure your model, or dangerous vulnerabilities found by your model, I kind of think you should not train the model in the first place.”

Anthropic upgraded its cybersecurity when it released Claude Opus 4 last May. The cyber measures included access controls and monitoring for the use of its models by outside parties, but Anthropic noted that it was not prepared to defend against sophisticated insiders or nation-state attackers. In February, the company said it was working on a “large, wide-reaching effort” to harden its internal systems, which it aimed to complete by July of next year.

Going forward, we’re willing to bet that Anthropic won’t repeat its apparent mistake in giving dozens of outside partners access to a powerful, unreleased model the same way again. If only to make nation-state attackers that want the model work a little harder!

Berkshire and Travelers began requesting for AI exclusions last fall and some states approved them to take effect as soon as the start of the year, the data show. Travelers and other insurers, such as AIG, said their subsidiaries requested the approvals based on AI-exclusion templates introduced last year by the Insurance Services Office, a private body that sets standards for the industry, and which was approved by the state regulators. But the companies said they have not implemented the exclusions. (A spokesperson for Berkshire’s insurance arm declined to comment, and spokespeople for Chubb did not respond to requests for comment.)

‘Unsustainable Risks’

AI exclusions from insurance policies could pose a problem for top AI developers and cloud providers that sell AI models to businesses. Those businesses could slow the release of AI-powered products or features until they sort out coverage for potential liabilities if the AI makes mistakes, said Connor Deeks, CEO of Codestrap, a boutique consultancy that advises several Fortune 500 companies on how to use AI.

“Companies are being pushed to use AI at unsustainable rates,” Deeks said. “The risks will become unsustainable and translate to financial losses.”

Most insurers haven’t yet taken steps to exclude AI from policies in California, New York, and Texas, where some of the most prominent AI providers are headquartered. That’s because the states haven’t yet approved ISO’s templates for AI exclusions, which insurers use. That means companies in those states could still rely on traditional policies for AI coverage.

The proposed exclusions would impact existing liability policies that theoretically cover a number of AI-related claims, such as employees who accuse businesses of using AI to discriminate against them or businesses that sell robots that damage property.

Insurers are monitoring recent consumer lawsuits against businesses using AI products and whether they will be forced to pay claims not intended to be covered under existing policies. Consumers filed about 800 such lawsuits in 2025, up 140% from the prior year, Wolfe Research said in a recent report.

“Insurers across many lines of coverage are really grappling with whether to draw a hard line on coverage for AI-driven losses or to leave it blurred,” said John Farley, a managing director of corporate insurance broker Gallagher.

General liability policies would only apply to some AI-related losses, experts said. Even if insurers cut out AI coverage from those policies, the companies they insure could ask them to cover AI losses under other policies, such as cybersecurity policies and errors and omissions policies that cover businesses making negligent mistakes like putting the wrong numbers in a customer contract.

But insurers’ proposed AI exclusions show they are likely to try to exclude AI from the other policies, according to Kevin Kalinich, Aon’s head of intangible assets, and Preet Gill, Lockton’s executive vice president of global technology risk. For instance, the Financial Times reported Wednesday that some insurers are moving to cap AI-related losses in cybersecurity policies.

To be sure, the largest companies could end up negotiating for AI coverage directly with their insurers, as they typically develop customized policies.

Insurers moving to cut AI coverage are probably trying to avoid a repeat of the 2010s, when a wave of cybersecurity attacks triggered corporate claims. Businesses in some cases successfully argued that traditional insurance policies covered losses from such attacks because they didn’t explicitly exclude them, brokerage executives said.

Insurers have mostly responded by excluding cybersecurity coverage from traditional policies and creating standalone cyber insurance—an approach they may repeat with generative AI.

Startups Pounce

Some firms such as Munich Re, the world’s largest reinsurer, and newer insurance startups have started selling policies to cover AI damages specifically.

Artificial Intelligence Underwriting Company, based in San Francisco, says it offers insurance policies to startups developing AI, such as voice AI maker ElevenLabs, covering losses of up to $50 million. Canadian upstart Armilla, meanwhile, says it has dozens of clients and covers up to $25 million in AI-related losses for AI developers and companies using AI. Insurance marketplace Lloyd’s of London is the primary risk-bearer of Armilla’s policies.

Other insurance startups such as Corgi, Mayflower Specialty and Embroker have also introduced standalone AI liability policies.

Corgi’s insurance policies are aimed at tech startups developing or using AI, covering losses related to AI mistakes as well as AI-related intellectual property infringements, property damage, or software service interruptions. Those policies typically cover up to $2 million in losses and cost anywhere from a few hundred dollars to several hundred thousand dollars a year.

“A lot of traditional carriers, they don’t know how to underwrite AI and they’re very spooked by AI,” said Emily Yuan, Corgi’s chief operating officer.

Traditional insurers such as Tokio Marine Holdings, W.R. Berkley, and Fairfax Financial began applying for AI exclusions in 2023 and 2024, when enterprise use of AI was still early. Then in the second half of 2025, the Insurance Services Office endorsed such moves and insurers have “meaningfully accelerated” efforts to get approvals for AI cutouts since then, said Wolfe Research analyst Tracy Benguigui in a recent report.

Tokio, Berkley and Fairfax Financial did not respond to requests for comment.

French weather service alerts police to tampering after suspicious Polymarket bets

Users of weather forum say Paris temperature data may have been manipulated for wagers on prediction market platform

France’s weather forecasting service has filed a police complaint after detecting anomalies in its temperature gauges at Paris-Charles de Gaulle airport, which coincided with a surge in well-timed bets on prediction market Polymarket.

Météo-France filed the complaint alleging interference with its equipment after temperatures measured at the airport spiked by several degrees Celsius in the space of a few minutes on April 6 and 15. The humidity also dropped sharply at the same time on April 15.

Users of weather forums who noticed the sudden movements said they may have been caused by people tampering with the equipment to shape the outcome of wagers on Polymarket. Multiple accounts on the platform, where traders can bet on real-world outcomes, appeared to have placed large bets on unexpected temperature rises in Paris.

Polymarket uses Météo-France data recorded at Charles de Gaulle to settle wagers on the highest temperature of the day in the city. Predicting weather patterns has become a popular activity on the platform.

On April 6, one wallet made $13,990 in profit on a stake of less than $30 after betting that the temperature in Paris would hit 21C. The account, which was opened this month, bought so-called event contracts when their price suggested the probability of a payout was just 0.2 per cent. The account’s bet was placed just before 7pm Paris time, which is when bets on that temperature increased.

At around the same time the account began betting on the temperature being higher than 18C, starting at an implied probability of 3.9 per cent, and won an additional $2,254.

On April 15, when the recorded temperature jumped in a few minutes from 18C to 22C before falling back, another wallet made more than $21,000 on a stake of just $119 by betting that the temperature that day would exceed 18C, at a time when the price of the contract suggested a probability of about 0.5 per cent.

On both days, trading volume on Polymarket’s “Highest temperature in Paris” market exceeded $500,000 — more than double the typical daily volume for this market.

Hunting for unusually confident or well-timed bets has become a flashpoint for prediction market traders on social media, although some warn that lucky or smart bets can be confused with market manipulation.

While users can track traders’ individual wallets and wagers on Polymarket, the crypto-based platform does not require most accounts on its international site to provide identification documents. This means that the company itself may not know who is behind a given wager.

“In light of physical findings on one of our instruments and the analysis of sensor data, Météo-France has indeed been led to file a complaint for interference with the operation of an automated data processing system with the Air Transport Gendarmerie Brigade in Roissy,” Météo-France said, declining to comment further. The complaint was first reported by broadcaster BFMTV.

Sébastien Brana, who runs online weather forum Infoclimat, whose members track Météo-France data almost in real time, said members had not immediately suspected wrongdoing when the first anomalies occurred on April 6.

“We thought it was an issue with the sensors . . . You can have sudden temperature changes at sundown when there is a storm as well. But the weather situation didn’t explain what was happening,” Brana said. “It became clear there was something else going on when it happened again on April 15.”

Météo-France has shared readings from its sensors since 2023, Brana said. 

Prediction markets, which allow customers to bet on binary outcomes of future events, have surged in popularity in recent years. However, the ability of users to make bets on highly specific real-world outcomes has triggered concerns that they are vulnerable to manipulation.

The FT reported last month that the US attack on Iran was preceded by a number of unusually large and well-timed bets, and found a similar pattern of bets in a market related to the US government’s capture of Venezuelan leader Nicolás Maduro in January.

Israel in February brought prosecutions against two reservists suspected of using classified information to bet on the country’s military operations.

Regulations around the use of prediction platforms differ between countries. In the UK, for example, the Gambling Commission considers Polymarket and its rival Kalshi, the biggest regulated platform in the US, to be unlicensed betting operators.

Polymarket, Paris police and airport police did not immediately respond to requests for comment.

Arnault warns Middle East war could spiral into ‘global catastrophe’

LVMH’s billionaire boss says luxury group’s recovery hinges on the conflict being resolved quickly

LVMH chief executive Bernard Arnault said the luxury giant’s recovery depended on a swift resolution of the war in the Middle East, which the billionaire warned could spiral into a “global catastrophe”.

“You will have noticed that the world is now in a very serious crisis in the Middle East . . . it’s very unpredictable,” Arnault said at LVMH’s annual meeting on Thursday.

LVMH’s chief said the group could return to growth across all its businesses this year — which range from fashion and handbags to hotels and high-end spirits — if the US and Israel’s war with Iran does not drag on.

“Otherwise, we will have to face a crisis,” Arnault said, adding that it could turn into a “global catastrophe with extremely serious and very negative economic developments. And at that point, who can say how things will turn out?”

The conflict, which threatens to drag down consumer spending around the world, has delayed an anticipated recovery in demand for luxury goods after a bruising couple of years.

The Paris-listed group behind Louis Vuitton and Dior said last week that the war knocked a percentage point off its first-quarter sales growth, which came in at 1 per cent on an organic basis. LVMH’s sales fell by as much as 70 per cent in early March, shortly after the outbreak of the war.

The US and Iran are currently embroiled in a stand-off in the Strait of Hormuz, causing major disruption to global shipping and energy markets. The countries have accused each other of breaching a ceasefire agreement in recent days and a fresh round of peace talks has yet to begin.

Arnault has cultivated a close relationship with US President Donald Trump, who he has known since the two worked in real estate in New York in the 1980s. He attended Trump’s inauguration in January last year.

LVMH shares have fallen 26 per cent so far this year. Shares in rivals Kering, Hermès and Richemont are all down by double digits in 2026. 

Toms Capital Investment Management demands $1.1tn pension and insurance group sells all or parts of its business

US asset manager Voya Financial is facing pressure from an activist investor to sell all or parts of the $1.1tn pension and insurance products empire, as a wave of consolidation sweeps the industry.

Toms Capital Investment Management, a US-based hedge fund, has built a position in Voya encouraging the company to consider putting itself up for sale or offloading its underperforming health insurer arm, according to people familiar with the matter.

The hedge fund believes Voya’s business insuring employers against health benefit claims is dragging down the wider company, the people said. The so-called stop-loss business ran a $10mn operating loss in the last quarter of 2025.

Voya, which was carved out of Dutch bank ING in 2014, mostly oversees higher-earning retirement and wealth management funds. Shares in Voya are roughly flat over the past two years, giving the company a market value of nearly $7bn on Thursday morning.

Voya has outperformed rivals on inflows, recently surpassing $1tn of assets across its wider retirement and investment platform, of which $360bn are actively managed. But TD Cowen analysts said in a note this month that the performance of its stop-loss business had “pressured sentiment”.

Voya has long been earmarked as a potential takeover target, as several large insurers with asset management arms have expressed a desire to add scale through dealmaking.

TCIM’s demands of Voya follow a sharp uptick in dealmaking in the asset management sector, as fund houses rush to scale up globally and private equity eyes business with transatlantic reach.

There were nearly $25bn worth of deals in the sector in the first three months of the year, more than half the total for all of last year, according to data provider Dealogic.

Last month, a consortium led by Nelson Peltz’s Trian Fund Management triumphed over Victory Capital in a bidding war for investment house Janus Henderson, sealing an $8.6bn deal. Earlier this year, UK asset manager Schroders was bought by rival Nuveen for £9.9bn.

Founded by alumni of London-based hedge fund GLG Partners in 2017, TCIM rarely launches public-facing activist campaigns and prefers to work behind the scenes. But TCIM has recently built stakes and pushed for strategic changes at Pringles maker Kellanova, US Steel, Tylenol maker Kenvue and retail giant Target.

Voya declined to comment, while TCIM did not immediately respond to a request for comment.