According to TechCrunch, the group exploited an external contractor's access and publicly available information; so far they have used the model for trivial tasks such as building test websites. TechCrunch ↗ An Anthropic spokesperson confirmed the incident, and the company is investigating it. In the Slovak media space, the topic was covered by the portal sector.sk. sector.sk ↗
Meanwhile, on April 17 Anthropic launched an entirely different product for paying customers. Claude Design promises to create prototypes, wireframes, mockups, pitch decks, and landing pages directly in chat, with a connection to the customer's repository or design files and further continuation via Claude Code into production code. TechCrunch ↗ The tool is available as a research preview for Pro, Max, Team, and Enterprise subscriptions. MacRumors ↗
The author of a day-long test on Medium sums up Claude Design by saying it "is not a Figma killer and isn't trying to be one" — it is useful for marketers, product managers, and small teams without a dedicated designer. Medium ↗ On the Slovak market, the tool received positive coverage from the portal TouchIT. TouchIT ↗ On the same day, Figma shares fell by 7.28%, according to Inbenta. However, the causality may not be straightforward: The Economic Times simultaneously reported on the departure of Anthropic founder Mike Krieger from Figma's board, which, together with the overall sentiment around AI design tools, may have contributed to the pressure on the stock. Inbenta ↗
OpenAI: Not Just Regulators, But Also a Molotov Cocktail
Slovak media wrote little about the attack on Sam Altman's home. According to police, 20-year-old Daniel Moreno-Gama from Texas threw a bottle with a burning rag at the metal gate of Altman's home in San Francisco around 4:00 a.m. on April 10, set it on fire, and fled. About an hour after this first incident, police arrested another suspect in front of OpenAI's headquarters, where, according to CNBC, he threatened to set the building on fire — authorities have not confirmed a connection between the two cases. CNBC ↗During the arrest, a document was found on Moreno-Gama in which, according to the criminal complaint, he spoke out against AI and AI company executives and wrote about the "impending extinction" of humanity. He faces several serious charges, including attempted murder and arson, in California state court. NPR ↗ BBC and CBS News add that he also faces federal charges for an illegal weapon and attempted destruction of property with an explosive. BBC ↗
A separate event is a shooting recorded two days later. Early Sunday morning, a shooting was reported near Altman's home; according to NBC Bay Area, police arrested 25-year-old Amanda Tom and 23-year-old Muhamad Tarik Hussein on suspicion of negligent discharge of a firearm. Investigators openly admit it is unclear whether these suspects were targeting Altman's residence at all, and the incident has not yet been officially linked to the April 10 attack. NBC Bay Area ↗
At the same time, civil and criminal pressure on OpenAI is escalating. A stalking victim from Silicon Valley filed a lawsuit on April 18, claiming that ChatGPT helped her ex-partner track her — this is the plaintiff's claim, not a verified fact. Fzone ↗
On April 21, the Florida prosecutor's office opened a criminal investigation into ChatGPT's role in a fatal shooting at a university in Tallahassee (we wrote about it →). At the same time, according to Fortune magazine, the company is supporting bill SB 3444 in Illinois, which, if certain conditions are met, would limit the legal liability of AI developers for so-called critical harms — defined as the death or serious injury of 100 or more people, damages exceeding one billion dollars, or misuse of the model in a chemical, biological, or nuclear weapon. The protection would apply only to companies that demonstrate they did not act intentionally or negligently and that publish safety reports. So this is not blanket immunity, as some of the more sensational headlines suggest. Interestingly, according to Fortune, Anthropic is actively opposing the same bill. Fortune ↗
The Slovak portal Aktuality covered the topic on April 22 with a considerably more sensational headline. Aktuality ↗ Against this backdrop, according to the Financial Times, some investors are questioning the high valuation estimates for OpenAI, which wire reports put in the range of hundreds of billions of dollars. iROZHLAS ↗
Four Countries, Four Different Responses
Last week brought an interesting contrast in how different governments are approaching AI. In the US, according to Axios, Anthropic outspent OpenAI on lobbying for the first time ever — in the first quarter of 2026 it spent $1.6 million compared to OpenAI's $1 million, four times the amount from a year earlier. Axios ↗ Anthropic's position is further complicated by a dispute with the Pentagon over its "acceptable use policy" — some members of the Trump administration have publicly labeled the company "woke."At the same time, according to the Washington Post, Maine's legislature has approved a moratorium on new data centers larger than 20 megawatts, running until November 2027; it is the first statewide restriction of its kind, but Governor Janet Mills has not yet signed it and is seeking an exemption for a specific $550 million project in the town of Jay. Washington Post ↗
Canada is heading in the opposite direction. The federal government launched the AI Sovereign Compute Infrastructure program, aimed at building domestic computing capacity for healthcare, energy, and research. Canada.ca ↗ At the same time, members of the governing Liberal Party approved a call for age limits on social media and AI chatbots at their convention in Montreal; the heritage minister, according to CBC, said the government is considering the proposal "very seriously." CBC ↗
New Zealand's outlook is a mix of optimism and lagging behind. During a visit to Auckland, Microsoft announced it would double its training program — aiming to retrain 200,000 New Zealanders by 2028. RNZ ↗ A critical commentary from NEWS WIRE points out that only 41% of New Zealand workers actually use AI at work, and speaks of the country's "closing window." NEWS WIRE ↗
Japan took the most ambitious step. On April 12, SoftBank, NEC, Honda, and Sony — together with Nippon Steel, Kobe Steel, and the banks MUFG, Sumitomo Mitsui, and Mizuho — founded a company called Japan AI Foundation Model Development. According to the Japan Times, the venture has 1 trillion yen in support available over five years, with the goal of building a physical AI model with approximately one trillion parameters. Japan Times ↗
TechWireAsia adds that SoftBank and NEC will lead the model's development, while Sony and Honda will deploy it in automotive, robotics, gaming, and semiconductors. TechWireAsia ↗ Japan's Ministry of Industry states its ambition to capture 30% of the global physical AI market by 2040 — a policy goal, not a prediction. Microsoft simultaneously announced $10 billion for AI infrastructure in Japan between 2026 and 2029. CNBC ↗
What to Watch
Three things are worth watching until summer. The first is the fate of Illinois SB 3444 — if it passes, it could become a template for other states and indirectly influence any continuation of Florida's criminal investigation into OpenAI.The second is whether Governor Mills will sign the moratorium in Maine or negotiate exemptions — signing it could embolden lawmakers in other states, where, according to the National Conference of State Legislatures, similar proposals are being considered in at least a dozen other legislatures.
And the third is whether the Japanese alliance built around SoftBank, Sony, Honda, and NEC will manage to produce a working prototype by the end of the year — this would be the first major attempt at AI sovereignty outside the USA-China-EU trio.
