TheChek News Flow

DAX:

MDAX:

SDAX:

The breaches, which follow a similar incident involving rival OpenAI and startup Hugging Face, were due to a mistake, Anthropic said

Anthropic said Thursday that software it was testing got onto the internet and hacked unsuspecting companies without the AI-maker’s knowledge in three separate incidents dating back to April.

The artificial-intelligence vendor didn’t say which companies had been hit, but all three were notified of the incident on Monday, Anthropic said.

The news comes one week after OpenAI revealed that its AI technology had broken out of a testing sandbox—a digital prison that was supposed to be disconnected from the internet—and hacked the AI company Hugging Face.

The OpenAI incident has rattled security researchers and AI professionals and acted as a reminder of the power and unpredictability of AI systems that act autonomously and are engineered to take a variety of actions to complete tasks and meet goals from users.

The incidents show that the AI companies need to have stronger, industrywide standards for isolating their systems during cyber testing, said Alex Stamos, chief product officer with the cybersecurity company Corridor.

But they also present a worrying glimpse at a future where ransomware criminals could conduct widespread attacks using increasingly capable AI systems, Stamos said. “We need to prepare for attackers to have these kinds of capabilities using open-weight models quite soon.”

The Anthropic hacks are sure to further stoke fears over the potential harms of powerful AI models and how to mitigate them. The White House has moved recently to increase its oversight of AI, while private industry has increased calls to preserve access to so-called open-weight models, which can be run on computer systems controlled by their users.

“AI is developing extremely fast with no real regulations to keep us safe,” Rep. Greg Casar (D., Texas) said last week.

After hearing about OpenAI’s problem, Anthropic decided to take a look at its own cyber tests to see if that had happened with any of its models.

After checking the logs of over 141,000 tests, the company discovered that Claude had indeed found its way onto the internet several times. But in the three hacks the company eventually discovered, Claude didn’t break out of a sandbox; it simply wandered out of systems where the sandbox didn’t exist.

According to Anthropic, “misconfiguration” on systems run by Anthropic and the company’s testing partner, the security firm Irregular, left the models with live internet access.

An Irregular spokeswoman said the company is investigating the incident.

The models had been told that the internet was unavailable, but during the course of testing, the models found their way online and hacked the companies using basic hacking techniques such as guessing weak passwords or finding their way onto systems that didn’t require authentication.

The models incorrectly believed that this hacking was part of their benchmarking exercise, Anthropic said.

Claude was able to get into a system at an unnamed security company by creating a malicious piece of software that was downloaded by the security company’s scanning system. That software then stole credentials that were used to get into infrastructure at the security company.

During this hack, Claude realized that what it was doing wasn’t something it should be doing in the real world, but it convinced itself that it was living in a simulation.

In the most serious case, Claude was trying to hack into a fictional company that just happened to share the same name as a real-world website. When Claude had a hard time breaking into its fake benchmarking company, it broke into a database belonging to the real-world company.

This database hack was the only instance where Claude kept hacking, even when it realized it had broken into a real-world company, Anthropic said.

In the two other cases, Claude either never realized that it had hacked a real-world company or it did figure out what was going on and then stopped hacking immediately.

The hacks started in April and involved Opus 4.7, Mythos 5 and an unnamed research model, Anthropic said.

The U.S. automaker could sell or spin off the business in its second-largest market over geopolitical concerns

Elon Musk designed Tesla’s TSLA 3.53%increase; up pointing triangle China business to be easily separated from its U.S. business because of geopolitical tensions. It might also come in handy if he proceeds with a SpaceX SPCX -0.31%decrease; down pointing triangle merger.

Musk in recent years instructed Tesla executives to organize the company with a “laser” between its U.S. and China businesses, according to people familiar with the planning. He wanted to ensure that in the event of geopolitical strife between the two countries, at least the U.S. half of Tesla would survive.

Now, some Tesla executives have been told to prepare for a separation of the China business ahead of a potential merger, a person familiar with those talks said. And Tesla advisers have discussed possible options for a separation, including a spinoff, sale or closure, another person said.

Any of those moves could have a profound impact on Tesla, whose Chinese operation transformed it into a consistently profitable global mass-market electric vehicle leader—and potentially impact its valuation if a SpaceX merger were to happen.

It is unclear how quickly Tesla could spin out or sell the China business, and it is possible that its plans could change.

Musk, who is chief executive of both Tesla and SpaceX, has talked up the crossover potential between both companies in recent months as he reorganizes them around artificial-intelligence projects. Investors and analysts have embraced the idea of a merger following a record-setting IPO for SpaceX, which raised $86 billion for the company in June.

“Obviously we can’t talk about, you know, combining companies and that kind of thing on earnings calls,” Musk told investors last week. “It is got to be done with the appropriate process.”

Musk was particularly concerned about Tesla’s dependence on China for battery cells and feared its access to semiconductors could be jeopardized if Beijing invaded Taiwan, the people familiar with the planning said. The goal was to be ready for 2026 or 2027 in anticipation of a war, they said.

Separating Tesla’s China operations—including an electric-vehicle factory there—would address potential conflicts arising from SpaceX’s work as a major U.S. defense contractor. Such a deal would aim to create a firewall between Tesla’s Chinese subsidiary and its U.S. business.

SpaceX faces significant restrictions due to its dealings with the U.S. government. The company is a major national-security defense contractor, with business lines that include launching top-secret satellites and operating internet services in war zones like Ukraine.

Sales to the U.S. government made up 20.9% of its business in 2025. That business is highly regulated, with the company restricted by export controls and, for some programs, classified government activities.

Executives have also discussed creating a separate sales entity to handle exports from Tesla’s Shanghai plant, one of the people said.

Tesla could create separate office systems and bar China-based employees’ direct access to other company units, the person said. While Tesla China already has some autonomy, employees more or less work together around the globe and the top executive in China, Tom Zhu, oversees all of Tesla’s global automotive business.

A potential merger between SpaceX and Tesla would likely trigger intense scrutiny from Beijing, given that a major U.S. defense contractor would control Tesla’s factories in China and the factories’ know-how and supply chain could be repurposed for the U.S. military, according to people familiar with the matter. Another concern for Beijing is that a deal could put data of China’s roughly two million Tesla owners in the hands of an American defense company, they said.

In the event of a merger, Beijing would likely seek Tesla’s commitment and business guardrails to prevent the influence of SpaceX on Tesla China and illicit diffusion of dual-use goods such as rare-earth materials from Tesla China to SpaceX, the people said.

Tesla has two major factories in Shanghai. They make electric cars and batteries that are sold in China and exported to various markets around the world, but not to the U.S. China is Tesla’s second-largest market after the U.S., making up around 18% of sales in the first half of 2026. Unlike many other Western competitors, Tesla’s automaking operation there isn’t part of a Chinese joint venture.

During the first Trump administration, Tesla launched an initiative called Project Carbon to relocate its Chinese suppliers to Mexico, a decision that went beyond concern over tariffs from the U.S. on Chinese goods, according to people involved in those plans.

Musk was particularly concerned about Tesla’s dependence on China for lithium iron phosphate, or LFP, battery cells, as well as its reliance on chips from Taiwan Semiconductor Manufacturing, the world’s biggest contract chip maker, which faces the threat of war between the island and Beijing, the people said.

In 2025, the EV maker decided that it would stop using China-based suppliers by 2027 in its U.S. factories, in part because of U.S. tariffs and government incentives for local production, The Wall Street Journal reported.

Mergers involving multinationals typically require regulatory approval across multiple jurisdictions, though most reviews focus primarily on antitrust concerns. In 2018, Qualcomm terminated its proposed purchase of Dutch chip maker NXP Semiconductors after failing to obtain clearance from China.

It is common for companies in the U.S. to have contingency plans for their China businesses and a number of U.S. companies have already spun out their China arms into independent companies. Starbucks recently sold a majority stake in its China business. In 2016, Yum Brands, owner of Pizza Hut and KFC, spun off its China unit and sold stakes to local investors.

Net profit climbed 32%

Sony 6758 0.55%increase; green up pointing triangle Group raised its annual earnings forecasts after first-quarter net profit came in stronger than expected, supported by a rise in earnings from its music and image-sensor businesses.

The Japanese electronics and entertainment company said Friday that net profit climbed 32% from a year earlier to 342.16 billion yen, equivalent to $2.15 billion, for the three months ended June. That beat the estimate of ¥262.6 billion in a poll of analysts by data provider Visible Alpha.

First-quarter revenue grew 8.2% to ¥2.838 trillion. Operating profit increased for its game, music, movie and image-sensor businesses.

For the fiscal year ending March 2027, it now expects revenue to increase 0.2% to ¥12.500 trillion and net profit to climb 17% to ¥1.210 trillion, driven partly by the strength of its entertainment and image-sensor businesses. The company previously projected revenue of ¥12.300 trillion and net profit of ¥1.160 trillion.

Sony Group has spent billions of dollars on acquisitions in recent years to beef up its entertainment content while relinquishing control of businesses in other areas.

In March, Sony roughly doubled its stake in the company that owns the Snoopy, Charlie Brown and other Peanuts characters, to 80% for about $460 million. It spun off its financial business in October.

The Japanese company said in May that it planned to form a joint venture with Taiwan Semiconductor Manufacturing Co. to develop and produce next-generation image sensors. Sony aims to limit capital expenditure with the move as it prepares for greater demand for “eyes” for machines in the age of artificial intelligence.

Sony on Thursday said it offered to acquire Japanese camera-lens maker Tamron, as part of efforts to strengthen its camera business. Sony is one of the largest shareholders of Tamron, holding a stake of about 15% as of the end of last year.

Situational Awareness Down 67% in July in AI Stock Rout

‘We let you down this month,’ Leopold Aschenbrenner’s hedge-fund firm told investors after selling bulk of stocks to Citadel

Leopold Aschenbrenner’s hedge-fund firm Situational Awareness is down around 67% so far in July after incurring heavy losses on AI stocks, according to a person who saw a letter the firm sent to investors Thursday.

“We let you down this month,” Aschenbrenner wrote in the investor letter, the contents of which were described to The Wall Street Journal.

The deep losses prompted the firm to sell the bulk of its stockholdings to Ken Griffin’s investment firm Citadel as it raced to gather cash to cover margin calls from its lenders.

Situational’s gains earlier in the year were so large that, even including July’s losses, the fund remains up about 80% on the year, the letter said. Through the end of May, Situational gained about 270%, the Journal previously reported.

Aschenbrenner party blamed short sellers who targeted the firm’s positions for exacerbating the fund’s losses, the letter said. The letter compared Situational’s experience to a bank run. Aschenbrenner told investors that the firm had removed all leverage from the portfolio.

The firm still holds a portfolio of investments in private companies including Anthropic.

Representatives for Situational Awareness didn’t immediately respond to requests for comment.

Situational had amassed well over $20 billion in assets under management since its founding around two years ago, making it one of the fastest-growing firms in years. Aschenbrenner had no professional investing experience when he started the firm with a few hundred million dollars.

He was seen by some as an artificial-intelligence oracle, with other investors closely tracking his firm’s movements as it placed big, leveraged bets on AI-related companies.

AI’s wunderkind is taken to the woodshed

The rise of AI has been swift and powerful.

As a result, there have been few cautionary tales, unlike in previous booms including the cable, shale and subprime bubbles that yielded Gary Winnick of Global Crossing, Chesapeake Energy’s Aubrey McClendon and Angelo Mozilo, the bronzed architect of subprime lender Countrywide Financial.

Now the rise and rise of AI has a poster child of the excesses creeping into what may be the hottest trade ever.

Leopold Aschenbrenner, a 24-year-old with no trading experience, over the past two years created a more than $20bn hedge fund sensation by loading up on AI bets.

His firm, called Situational Awareness, was up more than 400 per cent during the first six months of this year alone. Traders happily bid up each new position he revealed.

But on Wednesday, Aschenbrenner’s hot streak quickly ended as his fund, facing growing pressure from prime brokers and investors, sold off the majority of its public equity positions in a fire sale.

Ken Griffin’s hedge fund Citadel beat out rivals Millennium Management and Jane Street to buy a significant portion of Situational Awareness’s public stock holdings, which were once estimated at $16bn.

The rescue deal ranks among the largest and swiftest stock transactions in Wall Street history.

The transaction left Aschenbrenner with a slice of public stock holdings and takes in private companies such as Anthropic, the FT reveals in a deep dive into a 24-hour race involving titans of Wall Street.

After matriculating to Columbia University as a teenager, Aschenbrenner worked for the FTX Future Fund, the philanthropic arm of Sam Bankman-Fried’s empire before it collapsed in late 2022.

He later joined OpenAI’s “Superalignment” team before being fired over an alleged leak.

In 2024, he wrote a 165-page treatise on how AI would dramatically reshape society in the coming years, titled “Situational Awareness”. He capitalised on his newfound fame that year by raising a hedge fund with backing from former GitHub chief executive Nat Friedman and venture capitalist Peter Thiel. It quickly took off.

This July trouble began. Some of the hedge fund’s biggest positions appeared to be collapsing. The fund at one point held hundreds of millions of dollars in Bloom Energy and Sandisk, which both fell sharply.

Aschenbrenner wrote a letter to investors last week amid the AI market rout, in which he rattled off its wild returns — 439 per cent for the year, 1,551 per cent since its founding — and then gave a hint that trouble was brewing. 

While he said the firm had not been “immune” to the recent tech sell-off, he suggested that August 1 could be a great time to invest to take advantage of the “tremendously exciting” AI advances ahead.

As the situation worsened, trading giants began to explore deals to buy chunks of Situational Awareness’s portfolio.

Then Citadel, and its billionaire founder Griffin, entered the frame. He was “heavily involved” in the negotiations and at the final hour agreed to buy a significant portion of Situational Awareness’s public equity holdings.

“The real question is how did anyone invest so much money, and then lend so much money, to a [24-year-old] kid with no personal experience and no infrastructure,” said a person involved in the negotiations.