Anthropic is having a month.

The AI lab finished May by surpassing OpenAI in market share of business spending for the first time, Ramp just revealed. It raised $65 billion at a $965 billion valuation (also besting OpenAI) at the end of May, then waltzed into June by filing confidential paperwork for an IPO, reportedly on the strength of its first-ever profitable quarter.

Then on Friday, the Trump administration renewed its war on the model maker by sending a letter demanding it ban non-Americans, including Anthropic’s employees, from accessing its state-of-the-art models: the limited-release Mythos 5 and the more guarded version of Mythos released to the public three days earlier, called Fable 5.

This essentially forced Anthropic to pull its latest all-powerful model from the market altogether.

Although the White House invoked an obscure export control directive when ordering the ban, the exact cause remains unclear. The chatter was that hackers easily bypassed Fable 5’s guardrails, which were intended to prevent access to Mythos’ capabilities. That model is so good at finding security flaws in software code that Anthropic itself marketed it as dangerous and restricted its public release.

This new drama comes after Anthropic famously refused to allow the government to use its models for mass surveillance of Americans and fully autonomous weapons. As a result, in March, the Trump administration declared the company a supply-chain risk.

That didn’t deter Anthropic’s sales to businesses. Quite the opposite, Ramp’s data shows. Ironically, this latest feud with the Trump administration, which also appears to validate the hubbub over Mythos’ mythological power, may help rather than hurt Anthropic, according to Ramp’s lead economist, Ara Kharazian. Kharazian is the person who compiled the business-spending AI data.

“If anything, it’ll probably boost them,” Kharazian told TechCrunch. “Anthropic’s best month on record, as far as business adoption, was the month that the Department of Defense labeled them a supply-chain risk. There’s a lot of aura that comes with your model specifically being named too dangerous to use.”

Ramp’s data isn’t granular enough for us to see how much of a financial hit the company will take by pulling Mythos and Fable 5 off the market.

Still the data, from more than 70,000 businesses that use its platform, shows that customers heavily use Anthropic’s Opus models and that business use has been growing.

For instance, Ramp reported that Anthropic’s share of AI subscriptions paid for by businesses rose 2.5 percentage points in May to 41%. This compares to OpenAI, which commanded 39.5% of AI subscriptions by its customers, essentially flat from the prior month. (OpenAI still greatly leads Anthropic in overall consumer usage, according to new data from Sensor Tower.)

Beyond subscriptions, the vast majority of what companies spend money on is API calls to the model, which cover token use for activities like coding. Anthropic’s Claude Code has a strong reputation as a powerful AI coding tool.

Ramp can’t always see from the spending data which models most businesses are using. When it can see the model details — in about one-third of transactions — businesses are mostly spending on various flavors of Claude Opus, particularly the later versions. Opus is the model that preceded Mythos and is still openly available.

In fact, in late May, Anthropic released a new version, Opus 4.8.

Mythos had not been on the market for that long, having been released to limited users as of April. And Fable 5 was shut down after a few days.

While we can’t predict how this latest drama with the White House will impact Anthropic’s ability to go public as it hoped to (public-market investors tend to be wary of companies embroiled in controversies with the government), the numbers indicate that Anthropic’s available models are more popular with businesses than ever before.

Grassroots opposition to data center construction in the United States has reached a scale that is starting to reshape where and whether the AI industry can build. A new report from Data Center Watch, a tracker maintained by AI research firm 10a Labs, found that activists blocked or delayed at least 75 projects worth a combined $130 billion in the first quarter of 2026. According to NBC News, that is the most disruptions recorded in a three-month period since the group began tracking in 2023.

The pace represents a structural shift, not a spike. The total number and value of projects disrupted in Q1 roughly matched the full-year total for 2025, according to the report. The number of active anti-data center groups more than doubled from 396 at the end of 2025 to 833 by March, spread across 49 states, with Maryland, Ohio, and Texas hosting the most.

The opposition is bipartisan and locally driven. Communities are organising around electricity costs, water consumption, and noise, the same concerns that have already forced Denmark to pause all new grid connections for data centres and prompted the EU to ask households to cut peak electricity use because AI data centres are straining the grid.

Legislative momentum is building alongside the grassroots resistance. Data Center Watch counted 14 statewide measures introduced in Q1 2026, and a separate analysis by MultiState identified moratorium bills across 11 states with proposed pauses ranging from three months to four years. More than 300 data-center-related bills were introduced in statehouses in just the first six weeks of the year.

None of the statewide moratoriums have passed yet, but they are getting close. Maine’s legislature passed one in April that would have paused permitting for facilities drawing 20 megawatts or more, the first of its kind in the country. Governor Janet Mills vetoed it but said she would have signed it if the bill had exempted a specific project in Jay, Maine that had strong local support, and she separately signed a law barring data centers from state tax incentives.

A Heatmap Pro poll found that a majority of Americans would “strongly” oppose a data center being built near their home, a shift from a survey nine months earlier that showed the public roughly evenly divided. Gallup data puts the figure at 70% opposed. The speed of the opinion shift suggests the issue is crossing from local planning disputes into broader political territory.

The industry is spending as though the opposition will not hold. US utilities plan to spend $1.4 trillion by 2030 on grid infrastructure driven largely by data centre demand, and hyperscaler capital expenditure is projected to exceed $690 billion in 2026 alone. The gap between what the industry wants to build and what communities are willing to accept is widening faster than either side expected.

In some cases, opposition is now mobilising before any project is officially filed. The mere rumour of a data center has been enough to trigger organised resistance, according to the report. That pre-emptive organising makes siting decisions harder even in states without formal moratoriums, because local permitting bodies face political pressure before a single application lands on their desk.

The Atlantic published a contrarian essay on Friday arguing that the backlash is overblown and that data centers can bring real economic benefits to host communities. The piece acknowledged that opposing data centers is good politics but argued it is not always good policy. Whether that argument gains traction will depend on whether the industry can demonstrate tangible local benefits beyond tax revenue, something most communities have not yet seen.

The report paints a picture of an industry that assumed it could build its way through local opposition with money and speed, and a country that is deciding otherwise, one zoning board at a time.

South Korean authorities have imposed a record-breaking fine of $624 billion won (over $400 million) on retail giant Coupang after a data breach last year compromised the personal data of more than 34 million customers.

Seoul’s Personal Information Protection Commission issued the maximum penalty on Thursday following discovery of the breach in December 2025. The retail giant, which is headquartered in the U.S. but popular in South Korea and likened to the “Amazon of Asia,” had said the months-long data breach allowed a former employee to obtain names, email and shipping addresses, phone numbers, and order histories of about two-thirds of South Korea’s population.

Coupang told BBC News that it plans to challenge the regulator’s decision. The fine represents a rare case of a financial penalty issued against a U.S.-based firm. Korean lawmakers have accused some of their American counterparts of imposing political pressure after reports that U.S. representatives were linking the data breach with U.S.-South Korean bilateral ties in response to the case against Coupang’s executives.

U.S. companies rarely face financial sanctions or criminal prosecution for data breaches as a result of lacking laws and enforcement powers.

OpenAI has begun rolling out Lockdown Mode to ChatGPT, a new security setting designed to block attackers from stealing data through prompt injection attacks. The feature disables live web browsing, agent mode, deep research, image retrieval, Canvas networking, and file downloads. It is available to logged-in users across Free, Go, Plus, Pro, and self-serve ChatGPT Business plans.

Prompt injection remains what OpenAI calls a “frontier” problem affecting all large language models. The attack works by hiding malicious instructions in content the model processes, such as a webpage or uploaded file. If the model follows those instructions, it can be tricked into sending sensitive data to an attacker-controlled server.

Lockdown Mode does not stop injections from happening. A malicious payload embedded in a cached webpage or uploaded PDF can still influence the model’s behaviour. What it does is shut down the outbound pathways an attacker would use to exfiltrate the data. No live browsing means no network requests to external servers. No image retrieval means no pixel-based data channels.

“Lockdown Mode is designed to substantially reduce the risk of prompt injection-based data exfiltration, but it does not guarantee that data exfiltration cannot happen,” OpenAI said. “Risk may remain through enabled Apps, unforeseen combinations of capabilities, or newly discovered techniques.”

The trade-off is significant. With Lockdown Mode on, ChatGPT loses most of what makes its agent and research features useful. Live browsing drops to cached content only. Agent mode is gone entirely. Deep research is disabled. It is, as OpenAI acknowledges, “not intended for everyone.”

The feature arrives as prompt injection attacks on AI agents have become a growing concern. Security researchers have demonstrated hijacks against agents from Anthropic, Google, and Microsoft via their GitHub Actions integrations. All three paid bug bounties but published no public advisories. The underlying weakness is fundamental: LLMs cannot reliably separate data from instructions.

Lockdown Mode and Developer Mode cannot be used simultaneously. Turning one on disables the other. OpenAI also launched a separate session management feature that lets users review active ChatGPT sessions and log out of individual devices if they spot unauthorised activity.

The feature is a pragmatic concession. OpenAI is not claiming to have solved prompt injection. It is accepting that the problem persists and offering users a way to reduce their exposure by giving up functionality. For anyone handling sensitive data in ChatGPT, that trade-off is worth making. For everyone else, the expanding agent ecosystem and its growing attack surface mean the risk is only increasing.

OpenAI announced a new feature that it says will provide additional protection from prompt injection attacks, where malicious chatbot instructions are hidden in webpages and other content sources.

Among other things, Lockdown Mode will disable live web browsing (so you can only access cached content), the retrieval and display of images from the web (you can still generate images), deep research, and agent mode.

The company says that even with Lockdown Mode turned on, ChatGPT could still be vulnerable to prompt injections — which could, for example, “appear in cached web content or in an uploaded file, and could still affect the behavior or accuracy of a response.”

But the goal is to reduce the likelihood that sensitive data gets shared in the process.

“Lockdown Mode is not intended for everyone,” OpenAI says. “It is designed for people and organizations that handle sensitive data and want stricter protection from data exfiltration risks related to prompt injection.”

The company says it’s currently rolling Lockdown Mode out to self-serve ChatGPT Business accounts, as well as eligible personal accounts.

TL;DR

Blackstone-backed AirTrunk plans to invest $30 billion in India by 2030, building 5GW of data centre capacity across multiple states. The announcement comes six weeks after AirTrunk entered India through its acquisition of Lumina CloudInfra.

Six weeks ago, AirTrunk did not operate in India. Now it wants to spend $30 billion there.

The Blackstone-backed hyperscale data centre operator announced on Thursday that it plans to invest more than INR 3,000 billion ($30 billion) in India by 2030, building over 5 gigawatts of digital infrastructure capacity across multiple states and union territories. The figure represents planned spending, not committed capital, and the four-year timeline leaves considerable room for adjustment. Still, if executed, the programme would rank among the largest digital infrastructure commitments in the country’s history.

Prime Minister Narendra Modi publicly welcomed the commitment, saying it would strengthen India’s position as a global hub for cloud computing and AI. The endorsement followed meetings between AirTrunk founder and CEO Robin Khuda and federal and state government officials in Maharashtra and Andhra Pradesh.

From zero to $30 billion in six weeks

AirTrunk entered India in April through the acquisition of Lumina CloudInfra, which gave it a 600-megawatt development pipeline across Mumbai, Chennai, and Hyderabad. The new $30 billion plan represents a dramatic escalation of that position.

The centrepiece is a 3GW campus at the Raigad Penn Growth Centre on the outskirts of Mumbai, for which AirTrunk has signed a letter of intent for land allotment with the Maharashtra government. According to a single industry report, that project alone carries an estimated price tag of $21 billion, though the figure has not been confirmed by AirTrunk or the Maharashtra government.

“Capital is mobile, and India is creating the conditions for it to thrive,” Khuda said. “India is taking a top-down approach to AI with clear government-led initiatives, a world-class talent pool, and massive availability of renewable energy.”

Why India, why now

India’s data centre market has been accelerating since 2024, but the pace of new commitments in 2026 has been extraordinary. Google has pledged $15 billion for a southern Indian data centre hub. Microsoft has committed $17.5 billion. Amazon is targeting up to $35 billion by 2030. The Adani Group has reportedly outlined a $100 billion programme through 2035, including a 5GW renewable-powered hyperscale platform, though those figures come from industry reports rather than a formal company commitment.

The government has matched the private capital with policy. India’s February budget introduced a 20-year tax holiday through 2047 for foreign technology firms using Indian data centres for global cloud services. The IndiaAI Mission has received approximately £1 billion ($1.2 billion) in funding, and the India Semiconductor Mission has been backed with approximately £7.5 billion ($9 billion).

AI-related colocation leasing more than doubled to 348MW in the past year, now accounting for nearly 20% of total demand. Between March 2025 and April 2026, operators announced roughly 30 large projects adding about 3.5GW of planned capacity across the country. Schneider Electric expects its India data centre business to become its single largest unit within three to five years.

Blackstone’s hyperscale bet

AirTrunk is the vehicle through which Blackstone is making its largest infrastructure play in the Asia-Pacific region. The private equity giant acquired AirTrunk in December 2024 for an implied enterprise value of over A$24 billion ($16 billion), alongside Canada Pension Plan Investment Board, which took a 12% stake. It was the largest data centre transaction in history at the time.

Blackstone has since been expanding AirTrunk’s footprint aggressively. The platform now spans more than 3GW of operating and planned capacity across 20 campuses in six regions: Australia, Singapore, Japan, Malaysia, Hong Kong, and India. Separately, Blackstone is seeking up to $1.75 billion in a NYSE IPO for its Digital Infrastructure Trust, packaging hyperscaler-leased AI data centres as a public REIT.

The India push fits a clear pattern. Blackstone had already committed approximately $11 billion to Indian data centres through Lumina before the AirTrunk acquisition. The new $30 billion figure nearly triples that exposure.

The execution question

The numbers are staggering, but so is the gap between announcements and operational capacity. India’s total live IT capacity exceeded 1.6GW by the end of 2025, the product of years of cumulative buildout. Just 371MW was added in 2025 alone. AirTrunk’s proposed 5GW, combined with the commitments from Google, Microsoft, Amazon, and Adani, would require India to build more capacity in the next four years than it has built in its entire history, several times over.

The discussions between Khuda and government officials reportedly focused on precisely the bottlenecks that could slow that buildout: access to reliable and cost-effective power, renewable energy, sustainable water supply, talent development, streamlined approvals, and coordination between state and federal governments on strategic infrastructure.

India is not the only country chasing hyperscale AI infrastructure investment. Malaysia, Saudi Arabia, and several European nations are offering competing incentive packages. AirTrunk itself recently expanded its Malaysian platform to over 700MW. The $30 billion figure signals intent, but the timeline to 2030 leaves room for the kind of recalibration that large infrastructure programmes routinely undergo.

What is not in question is the direction of travel. Whether the final number is $30 billion or something smaller, India is rapidly becoming one of the world’s primary construction sites for the physical infrastructure that AI requires. The question is whether the country’s grid, water supply, and planning systems can keep pace with the capital flooding in.

A former IBM cybersecurity executive accused the company of getting hacked three times in the previous decade by foreign governments and then covering up the breaches. 

In a lawsuit unsealed this week but filed in 2020, William Barlow, who was IBM’s vice president of threat intelligence until August 2019, said IBM concluded Chinese hackers breached its core network between 2013 and 2016 but that the company then covered up the breaches and never disclosed them. Barlow also said at least two IBM subsidiaries were also breached, and that IBM covered up those breaches as well.

Barlow alleged in his complaint that IBM’s core network was “routinely hacked by foreign state actors and others,” adding that data was frequently stolen and government agencies were “never notified.” 

While the alleged breaches date back more than a decade, the news shows that cyberattacks, even those affecting large public tech companies such as IBM, sometimes never get disclosed, either to the public or to relevant government authorities. IBM is a major cybersecurity vendor to the U.S. federal government, which makes the alleged concealment especially significant. In the last few years, several data breach notification laws have been passed to counter this problem.   

Bloomberg first reported on the lawsuit.

IBM spokesperson Miki Carver declined to answer specific questions about the lawsuit and the underlying accusations. Instead, Carver told TechCrunch, “This complaint was filed six years ago, and the U.S. Department of Justice declined to intervene. IBM is confident that our actions followed the letter of the law.”

In particular, Barlow said IBM was among several victims of a hacking campaign carried out by APT 10, a Chinese government-linked group that then-FBI Director Christopher Wray said had targeted a “Who’s Who” of the global economy when its members were indicted in 2018. The hackers broke into both the company’s network and the data it maintained there in partnership with AT&T. 

Barlow alleged that in March 2017, intelligence officials from Australia, Canada, New Zealand, United States, and the United Kingdom — the so-called Five Eyes alliance — warned IBM of the breach, which prompted an internal investigation.

According to the complaint, the investigation concluded that APT 10 potentially breached IBM’s network more than 56,000 times between 2013 and 2016. Crucially, the company said it could not investigate further because it had not kept logs of who accessed its network and when — a basic security practice.

IBM then allegedly failed to alert any authorities or the U.S. government, one of its main customers. 

“As IBM and AT&T’s Core Networks’ infrastructure is archaic, hackers have been able to gain access to the system on numerous occasions and can roam almost anywhere undetected,” read the complaint, which explained that IBM’s internal investigation concluded four servers were compromised in the APT 10 hacking campaign.

“The attackers have compromised and/or accessed nearly 400 compromised accounts and almost 200 total systems and servers across every IBM business unit, eighteen countries, and multiple IBM products,” said an internal IBM report about the investigation into the breach, according to the complaint.

Jason Brown, a lawyer representing Barlow, told TechCrunch that his firm is “looking forward to aggressively litigating the matter.” 

“You can’t sell cybersecurity to the federal government while allegedly having these security problems within your own company,” said Brown. 

According to Barlow, other breaches he was aware of affected Trusteer, a cybersecurity startup acquired by IBM in 2013, which he says was breached in 2018; and Truven, a healthcare data startup IBM acquired in 2016, which he says was breached multiple times after the acquisition.

In both cases, Barlow accused IBM of failing to properly investigate and disclose these breaches. 

Environmental activist Erin Brockovich has a new mission: bringing more transparency to data center construction and the impact those data centers have on nearby communities.

Brockovich — who was famously played by Julia Roberts in a film dramatizing her legal case against Pacific Gas & Electric — recently launched a website with a map of data centers across the United States.

The website describes the map as “a work in progress” that includes data centers reported by members of the surrounding community. In a Substack post, Brockovich said that after putting out a call for reports of data center-related issues in April, she received nearly 4,000 submissions in the first month alone.

“The single most common concern — more than noise, more than water usage, more than rising utility bills — is the one word that keeps appearing in submission after submission: transparency,” she wrote.

Brockovich added that she’s not “making a blanket argument against data centers” or AI, but rather against “the pattern our map documents: projects announced after permits are already secured, developers who don’t return calls, local officials who signed NDAs before their neighbors knew a project was being considered.”