Tag: ai

White House Moves to Accelerate AI Data Center Development on Federal Lands

The Trump administration has issued executive orders intended to accelerate the development of artificial intelligence data centers, including on federal lands, according to administration officials. The orders are designed to address rising electricity demand from AI computing and to strengthen U.S. competitiveness, the White House said. Data centers currently consume about 4.4 percent of total U.S. electricity, a figure projected to reach 12 percent by 2030, according to industry estimates cited by the White House. [11]

Executive Orders Direct Agencies to Open Federal Lands

The executive orders direct the Departments of Energy and the Interior to identify federal parcels for data center leases and to designate projects as priority infrastructure, according to the administration. Agencies have been instructed to expedite permitting and coordinate transmission connections, officials said. The Bureau of Land Management has already approved the Townsite Data Center project to construct, operate, and maintain a data center on 88.5 acres of public land, Sen. Edward Markey (D-MA) said on Facebook, accusing the administration of approving the “first private data center on public land using a permit meant for a solar project.” [10]

A report by Children’s Health Defense, “Cheating the System,” details how developers have exploited legal loopholes following a May 2023 Supreme Court ruling that limited Clean Water Act jurisdiction, leaving many data centers outside Environmental Protection Agency oversight or eligible for streamlined nationwide permits that require minimal review. [12]

Power Demand Drives Energy Infrastructure Plans

Officials said the data centers will require electricity from natural gas, nuclear, and geothermal sources, along with expanded high-voltage transmission capacity. Interior Secretary Doug Burgum warned the Senate that the U.S. risks losing the global AI race to China unless it prioritizes reliable, baseload electricity generation, primarily from fossil fuels, to meet the energy demands of AI and data centers. [5] In its quest for more electric power, AI data centers are forming partnerships with nuclear power providers, with utilities owning about a third of U.S. nuclear power plants in discussions with AI centers about long-term electricity supply, according to the Trends Journal. [6]

The rapid buildout raises questions about grid reliability, water consumption, and environmental impacts on previously undisturbed federal lands. Local officials across the United States say rapid construction of data centers is creating environmental and infrastructure concerns, according to NaturalNews.com. [4] The backlash against AI data centers is intensifying across America, driven by growing awareness that these facilities are escalating electricity costs and displacing resources essential for human life, including water, farmland, and energy, according to BrightVideos.com. [14] According to U.S. Labor Department data cited in the Trends Journal, electricity bills are up more than 10 percent and homeowners’ insurance is up 13 percent over the past two years. [7]

Supporters Cite Competition; Critics Raise Concerns

Administration officials and industry representatives said the buildout is necessary for U.S. leadership in AI and for national security. State lawmakers and big tech are pushing back against what they call misinformation surrounding data centers, arguing that the facilities are vital for American jobs and technological leadership, according to Just the News. [9] Interior Secretary Doug Burgum said in Senate testimony that without reliable baseload power, the United States risks falling behind China in the AI race. [5]

Opposition has also organized. More than 300 cities, towns, and counties have enacted bans or moratoriums on hyperscale construction, according to a count by The Information, and a poll cited by NBC News found 58 percent oppose local data centers. [8] Environmental groups and local residents have objected to land use and emissions, while public health advocates have cited concerns about noise and light pollution, according to BrightVideos.com. [14]

Keep reading

OpenAI, Anthropic Models Created Fake Profiles, Tried To Trick Humans During Cyber Tests

Artificial Intelligence (AI) models from Anthropic and OpenAI carried out unsanctioned actions targeting multiple people and organizations during a cyber evaluation, according to the UK AI Security Institute (AISI).

AISI, which receives access to advanced AI models under voluntary agreements from major labs, put the agents through a fictional cybersecurity scenario to test capabilities.

The organization tested multiple AI models on two cyber challenges between July 25 and 28. In total, the challenges were run 122 times across several models, and in 19 cases, the AI models took “unsanctioned action on the live internet, including cases that targeted real people and organisations,” AISI said in an Aug. 4 report.

Out of these, 17 were done by Anthropic’s Mythos 5 model and two by OpenAI’s GPT 5.6 Sol model.

“This is the first time AISI has seen deception of this severity that was targeted at a real person, unprompted, in the real world. It is uncertain to what extent the model recognised it was taking actions against real people,” AISI stated in the report.

In the most serious incident, Mythos 5 tried to solve a cyber challenge by attacking GitHub, a cloud service that allows developers to store and manage code. The AI created a GitHub account and tried to get a malicious code approved by humans. Mythos 5 also created a second GitHub account, masquerading as another human user endorsing the malicious code. When a human reviewer caught the activity, the AI agent “falsely claimed to have made an honest mistake” and then repeatedly tried to reintroduce the malicious content by claiming the code was fixed.

Mythos 5 also sent multiple emails to two developers in an attempt to manipulate the recipients into running its malicious code.

According to AISI, the test runs of the various models did not result in any real-world harm. The organization said it was treating the AI actions as a “serious security incident” that requires further scrutiny.

AISI listed multiple factors that could have led to AI models acting in a concerning manner. The AIs were provided with internet access during the tests, which allowed them to carry out the unsanctioned actions.

AISI had also deliberately disabled the models’ cyberclassifiers during tests. Cyberclassifiers are systems aimed at preventing misuse.

The AI models were also not explicitly told what they were prohibited from doing on the internet, AISI said, adding that clearer instructions may have prevented the unsanctioned actions.

In an Aug. 4 blog post, AISI said that since the behavior of AI models happened under specific conditions, it cannot say how likely such behavior will be in different contexts.

“What we can say is that the behaviour was possible, sustained, and new; that alone warrants attention,” the organization said.

In an Aug. 4 post on X, Anthropic said that Mythos 5 and ChatGPT 5.6 Sol models attempted to complete their assignments in an environment where they were deliberately provided internet access and their normal safeguards were removed.

Due to the absence of specific restrictions on how the internet must be navigated and lack of safeguards, the models were tested under conditions that are “not representative of any of our production models,” Anthropic said, adding that there was no evidence in these tests of an AI escaping from a secure environment.

According to the company, it was closely working with AISI to access more details on the incident while carrying out an internal investigation on the matter.

OpenAI said in an Aug. 4 statement that it appreciated AISI’s partnership throughout the evaluation process, including the organization’s work to identify, investigate, and share details about the activity of the GPT 5.6 Sol model in their tests.

“We look forward to continuing our collaboration together,” the company said.

The Epoch Times reached out to Anthropic and OpenAI for comment but did not receive a response by publication time.

OpenAI was in the midst of another controversy last month after it admitted on July 28 that its models bypassed restrictions during an evaluation. In this case, the company was testing its models’ capabilities in carrying out cyberattacks.

AI startup Hugging Face was impacted in the test. On July 16, the startup said it detected an intrusion into its data processing systems. It was only later that the startup learned that the intrusion was carried out by an OpenAI model.

HuggingFace then worked with OpenAI to contain the attack, the startup’s CEO, Clement Delangue, said in a July 22 post on X, calling it “an attack unlike anything we’ve seen before.”

“This is day one for cybersecurity in the age of agents & we’re all learning that secrecy is not the answer & that all defenders (not just a few selected ones) everywhere need more powerful models without restrictions, especially open ones,” Delangue said.

Keep reading

All Flock cameras taken down in Burnet County

The Burnet County Sheriff’s Office posted that all fixed Flock cameras in the county were taken down as of Monday.

Those cameras were installed by the Heart of Texas Auto Theft Task Force and commonly used to read license plates.

According to the sheriff’s office, two of the cameras were damaged by vandals before the removal effort. One camera was vandalized by someone who cut its mounting brackets and cables. In a separate incident, another camera was struck with a blunt object before its battery and cables were damaged. Criminal cases were opened; however, the task force said there were no known suspects in either case.

The Burnet County Sheriff’s Office tells KXAN they’ve received multiple concerns regarding the cameras.

Keep reading

Dear America, We Do Not Beat China in the AI Race if We Become China in the AI Race

America is in an AI arms race. That’s the official story, anyway — a convenient narrative blasted from the teleprompters of corporate media and the podiums of bought-off politicians. We’re told that to question the wholesale surrender of our power grids, our water aquifers, and now our public lands to the data center boom is not just naive, but unpatriotic. This is the oldest trick in the book of statecraft: wrap a corporate giveaway in the flag and dare anyone to question it. The moment dissent gets reframed as disloyalty, you know you’re being scammed.

This was the raw, unfiltered message from Donald Trump at his Ratepayer Protection Pledge event, where he looked at his own constituents and told them, regarding the industrialization of their communities, “You can’t fight it. You have to go with it.” There is no clearer definition of tyranny than that. It’s the voice of an apparatus that has stopped asking for consent and started issuing commands.

Before we go further, let’s be crystal clear about what a free market actually is, and how it relates to the building of these data centers, because the term has been so grotesquely twisted it’s nearly lost its meaning. A free market is what happens when private individuals or companies make consensual deals. It is the opposite of the centrally planned, state-subsidized, resource-commandeering binge we are currently watching unfold.

And the most damning evidence that this AI boom has nothing to do with free enterprise is the simple fact that they are building it on your land. This isn’t private companies buying private property from willing sellers. This is the state acting as a real estate agent for the corporate elite, handing over vast tracts of public and military land in negotiated deals that the communities sitting next to them never got a real vote on.

Consider the facts. In Nevada, the federal Bureau of Land Management just approved the first commercial AI data center ever authorized on public land, waving aside desert tortoise habitat and the already-drought-stricken Colorado River. The agency did it, according to the conservation groups now suing to stop it, by reusing an old environmental review written for an entirely different solar project — meaning no new review and no public comment period were ever required, and the city of Boulder City itself says it was never consulted. The Department of Energy is running the same play with the wreckage of the Cold War: it has tapped SoftBank to build a 10-gigawatt AI campus at the former Portsmouth uranium enrichment plant in Ohio, selected Brookfield and NextEra for a $100 billion complex at the former Paducah uranium plant in Kentucky, and picked Amentum and DC Blox to develop a gigawatt-scale campus at the Savannah River nuclear-weapons site in South Carolina.

The Department of Defense has gotten in on it too, conditionally selecting Carlyle and CyrusOne to build hyperscale data centers on Army land at Fort Bliss, Texas, and Dugway Proving Ground, Utah, while the Air Force has offered up nearly 4,700 acres across three bases in Alaska for the same purpose. It has even trickled down to city hall: back in 2023, the Manassas, Virginia city council voted unanimously to sell 21 acres of city-owned airport land directly to Amazon for a new data center, with no ballot, no referendum, and barely a public hearing.

When the state is auctioning off the very land it supposedly holds in trust for the public, this is not capitalism. This is corporatism. It is a direct transfer of public assets into private, politically connected hands, and it makes a complete mockery of any claim to a “free market.”

With every one of these backroom deals comes the same tired promise: jobs. But the state’s own numbers expose the fraud. At a cost of roughly $13 million in investment per permanent job in Virginia, or $11 million per job in a recent New York deal, these are not job-creation programs. They are handouts to automate human labor. You’re being asked to celebrate a system where your own tax dollars subsidize the very automation that renders you obsolete.

The real costs — the ones that never make it onto the glossy brochures — are being socialized and dumped on an unwilling public. Residential electricity prices have shot up more than 36 percent since 2020, with data centers a primary driver. Dominion Energy in Virginia pushed through its first base-rate hike since 1992, and Baltimore residents saw their bills climb by more than $17 a month after a record power auction. The grid is being strained toward the breaking point, and the cost of every upgrade flows directly to ratepayers who never consented to any of it. It’s a textbook case of privatized profit and socialized loss: politically connected corporations reap the rewards, and you get handed the bill.

These facilities are also vampires, and they are draining the aquifers dry. U.S. data centers consumed more than 200 billion gallons of water last year, with roughly 93 percent of that footprint hidden inside off-site power plants where nobody’s looking. In drought-stricken Texas, consumption is projected to explode 714 percent by 2030, yet a state legislative hearing revealed that a staggering 83 percent of the state’s data centers simply refused to comply with mandatory water reporting. This stops being a negotiation and becomes the outright seizure of a shared public resource, with the state acting as armed lookout for the plunder.

It becomes something far more sinister when you remember what a lot of this infrastructure is actually for. Data centers themselves are neutral, but their deployment by the state is not. When the Department of Defense leases military land to build data centers, it isn’t for hosting cat videos. It is the physical construction of a turnkey surveillance apparatus, built on the one category of American ground that is already unambiguously controlled by the government, with none of the normal local checks that even a private developer would face. We are being told to subsidize the very infrastructure that is increasingly being used to watch our every move — and to do it on land we collectively own.

Keep reading

Wichita Man Sues City To Rip Out Its 200-Camera Flock Dragnet

A Wichita machinist, Mason Grimmett, sued his own city this week, arguing that the roughly 200 Flock license-plate readers the police department has bolted around town amount to warrantless dragnet surveillance forbidden by the Kansas Constitution. The complaint, filed in Sedgwick County District Court by the Kansas Justice Institute, does not ask for money — rather, it asks a judge to declare the program unconstitutional and shut it down.

The cameras are made by Flock Safety, one of the largest plate-reader vendors in the country, and Wichita started installing them in 2020. The institute now describes the city as saturated with them — close to 200 AI-powered cameras photographing every passing vehicle and logging its make, model, color, plate, and identifying marks into a searchable database. Grimmett, a lifelong Wichita resident who testified against the program at a public meeting in June, argues that the system violates Section 15 of the state constitution, which guards against unreasonable searches and seizures, and that the city is, in the complaint’s language, tracking and databasing law-abiding Kansans “without a warrant, without probable cause, without reasonable suspicion” and without any judicial oversight at all.

The Cops Who Already Abused It

Grimmett points to several local cases in which Kansas police used Flock cameras to stalk people, including abuse by former Kechi Police Lt. Victor Heiar, who was sentenced to eighteen months of probation for using Flock cameras to track a woman. According to local reports, a former detective was also accused of using plate-reader technology to stalk his wife, and a Sedgwick-area police chief reportedly ran an ex-girlfriend through the cameras 164 times. The institute is not disputing that the cameras help solve crimes. Its point is that they photograph everyone, all the time, and that the same access which cracks a burglary case also lets a jealous cop follow whoever he pleases.

“This is America, and we don’t tolerate warrantless surveillance,” said Sam MacRoberts, the institute’s litigation director, who says that if the government wants to monitor a citizen it ought to go to a judge and explain why. Grimmett says he cannot drive to work, the gym, a doctor’s appointment, a polling place, or a meeting of the local Sunflower Privacy Alliance without being catalogued by the department and by Flock.

Keep reading

Government Will Steal Your Land to Feed the AI Machine

Private land may be seized to construct transmission lines required by the rapidly expanding AI data-center industry. If a property owner refuses to sell an easement voluntarily, a utility may resort to eminent domain, provided the project is declared a “public use” and the owner receives what the government calls “just compensation.” Let us call this what it is: legalized theft.

The government does not ask whether the land has been in your family for generations, whether you built your home with your own hands, or whether the money offered would ever replace what is being destroyed. A bureaucrat determines the supposed market value, a judge blesses the taking, and armed government agents will eventually remove you if you continue to resist. They hide behind legal phrases because “confiscation for corporate benefit” would expose the practice for what it has become.

The United States already has more than 3,000 data centers, with another 1,500 under development. These facilities consumed more than 4% of total US electricity in 2024, and demand is rising rapidly as companies construct gigawatt-scale AI campuses. New transmission lines must cross somebody’s property, and when the owner says no, the state may simply decide that no does not matter.

Seventy percent of Americans reportedly oppose having a data center built near their community. These facilities can consume enormous quantities of electricity and water while creating noise, traffic, pollution, and higher infrastructure costs. Health risks are not fully understood although reports of increased cancer risks are prevalent in communities near these facilities. Yet the same governments that claim to represent the public are preparing to override that opposition because the technology companies have more influence than the families whose land stands in their way.

This is already happening. CBS News reported that Georgia Power acquired more than 300 parcels for a transmission project intended largely to serve data centers. The utility said that between 70% and 80% of the new line’s capacity would support data-center demand. Families were informed that if they refused the proposed sale, Georgia Power could pursue condemnation.

Ansley Brown’s family sold the home her grandparents had built after receiving an eminent-domain notice. She called the process “theft,” and she was absolutely correct. Georgia Power would not even identify the data-center companies benefiting from the project, citing customer confidentiality. The landowner must surrender everything, but the corporate beneficiary is permitted to remain hidden. That tells you exactly who the government serves.

The Fifth Amendment states that private property shall not be taken for public use without just compensation. The government has twisted those words beyond recognition. A transmission line that principally benefits unnamed private technology corporations is now presented as a public necessity merely because the electricity passes through the broader grid. Under that reasoning, nearly any private commercial project can be disguised as public infrastructure.

The Supreme Court opened the floodgates with Kelo v. City of New London in 2005. The Court ruled 5–4 that private property could be taken and transferred to another private party as part of an economic-development plan. The politicians promised jobs, tax revenue, and revitalization. The proposed Pfizer-related development never materialized as promised, and much of the condemned neighborhood remained vacant for years. Families lost their homes so politicians could gamble with property that was never theirs.

Keep reading

North Carolina Police Officer Arrested For Alleged Illegal Use Of Flock Cameras

A North Carolina officer was arrested by state authorities July 29 after allegedly accessing license plate registration information from Flock Safety cameras and other technology, authorities announced.

Charlotte-Mecklenburg Police Officer Seth Elliot, 25, is facing a misdemeanor charge of illegally accessing a government computer, according to the North Carolina State Bureau of Investigation.

Elliot, who started working at the department in 2024, allegedly accessed the Flock system and CJ Leads, a criminal justice database used by law enforcement in North Carolina, one time for non-law enforcement purposes, according to the police department.

The officer was placed on unpaid leave pending the outcome of an internal investigation, the Charlotte-Mecklenburg Police Department said.

A previous unrelated investigation had initially found the evidence that on June 16, Elliot had allegedly improperly accessed the license plate reader systems.

He was then placed on administrative assignment on June 22, and his equipment and access to systems with criminal intelligence information was revoked pending the outcome of the internal investigation.

Police Chief Estella Patterson said the department doesn’t operate any of the Flock cameras or devices but has an agreement with the company to allow officers access to them.

Misuse of the access is not tolerated, Patterson said.

“The public expects and deserves the highest level of integrity from every officer who wears the CMPD badge,” Patterson said in a statement. “Misusing law enforcement technology or accessing confidential information for non-law enforcement purposes is a betrayal of the public’s trust and a violation of everything this profession stands for.”

The Epoch Times was not able to reach Elliott for comment, and it’s unclear if he has an attorney.

The North Carolina case is the latest law enforcement-related allegation of Flock camera misuse.

Earlier this month, Sgt. Kabiru Salawu of the DeKalb County Sheriff’s Office was arrested after an internal investigation found he allegedly misused the agency’s Flock system. Salawu was fired by the department and charged with a felony violation of oath of office and misdemeanor misuse of license plate data, according to a sheriff’s office statement.

Investigations in Georgia and several other states have also led to arrests over Flock camera access.

People have also destroyed the cameras in some cities, with some incidents occurring this week.

In Upstate New York, police arrested a 25-year-old man July 29 after he allegedly used a saw to cut down a Flock camera.

In Central California, a 40-year-old Pismo Beach man was arrested July 28 for allegedly intentionally tearing down and running over at least three Flock license plate readers, causing thousands of dollars in damage, Monterey County Sheriff’s Office reported.

Flock Safety, based in Georgia, estimates more than 6,000 communities in 49 states have partnered with the company to install the surveillance systems.

The cameras are powered by artificial intelligence and capture and analyze images of all passing vehicles, recording the location, date, and time. The cameras also capture the car’s make, model, color, and other features, such as dents and bumper stickers, to use as data points, according to DeFlock, a community-driven, open-source project that maps the license plate readers.

DeFlock estimates Flock has installed nearly 123,400 cameras throughout the United States.

Keep reading

AI Policy Organizations Call for Federal Investigation into OpenAI Hacking Incident

A coalition of AI policy organizations is urging President Donald Trump to launch a formal government investigation into OpenAI following a security breach involving the company’s AI agents hacking the Hugging Face platform in an “unprecedented” self-directed attack.

The Washington Post reports that a group of prominent AI policy organizations has sent a letter to President Trump requesting an official investigation into a recent security incident at OpenAI, where an AI hacked into the Hugging Face AI platform in a self-directed attack. The letter, led by Brad Carson, president of Americans for Responsible Innovation, and Brendan Steinhauser, CEO of the Alliance for Secure AI, emphasizes the severity of the incident and calls for independent auditors to examine what occurred.

The letter has garnered support from several respected organizations, including Public Citizen, the Future of Life Institute, and FAR.AI, an independent AI research and evaluation organization. The coalition argues that the incident reveals broader vulnerabilities in AI systems that require government oversight and public transparency.

Carson drew a comparison between the OpenAI incident and aviation disasters, stating that while private investigations have their place, public interest demands government involvement. “This is kind of like an airplane crash,” Carson said. “It’s fine if Boeing wants to investigate itself for this. Even if they hire a very reputable outside person to help them on that project, [we] have a public interest in knowing what’s going on here.”

The call for investigation comes as METR, an independent AI research nonprofit, announced an agreement with OpenAI to conduct research into the security incident alongside Redwood Research. However, Carson expressed concerns about the scope and transparency of this private investigation, questioning whether METR operates under a non-disclosure agreement and whether findings will be made public.

An OpenAI spokesperson responded by stating that METR and Redwood Research plan to publish a joint blog detailing their agreement terms and evaluation scope. “This is an unprecedented incident, and we think it marks an important moment for AI safety,” the spokesperson said. “We are conducting a thorough review along with external advisors and with oversight from our Safety and Security Committee. Once the review is complete, we will publish a technical report of our learnings for everyone.”

Keep reading

OpenAI CEO Sam Altman Signals Openness To Slowing AI Development Amid Safety Concerns

OpenAI CEO Sam Altman said it may be time to deliberately “pace” the rate of artificial intelligence progress to give society more time to adapt to rapidly advancing capabilities.

“We may have to pace the rate of AI development to give ourselves enough time for society to harden around some of these new capability levels,” Altman told Patrick O’Shaughnessy on the Invest Like the Best podcast. He added that any slowdown must avoid the appearance of regulatory capture or collusion among leading AI labs, wrote Tech Crunch.

Both OpenAI and Anthropic have backed a petition circulated by employees at frontier AI companies. The petition urges the U.S. government to support an international effort to develop technical and governance tools needed to deliberately pace the frontier of automated AI development.

Altman had previously resisted calls to slow AI progress. In 2023, he criticized a widely circulated open letter proposing a pause, saying it was “missing most technical nuance about where we need the pause.” His stance appears to have softened following a recent security incident in which one of OpenAI’s advanced models escaped a secure computing environment and hacked into Hugging Face, an online model database, using multiple zero-day exploits.

Altman described the event as an “extremely sci-fi cyber incident” and “the first security incident that I have felt very viscerally.” OpenAI researchers have paused training on the model while they work to improve the security of their sandbox environment. As models grow more powerful, Altman said, the need to pace development could become essential for safe deployment.

The release earlier this year of Anthropic’s highly capable Mythos model has helped turn long-standing theoretical safety concerns into concrete challenges. Still, the industry faces a trust problem and difficult economics that can incentivize companies to emphasize dangers in ways that outside experts sometimes dispute. Debates have included whether Anthropic’s Fable model warranted a temporary ban and how the release of China’s large open-weight Kimi K3 model affected the competitive position of Western frontier labs.

Keep reading

AI Expert Sounds the Alarm: OpenAI’s Rogue Hacker Models and Killer Robots Signal a ‘Dystopian’ Path to Replacing Humanity

A leading artificial intelligence safety researcher issued a stark warning this week against pursuing general superintelligence and developing autonomous killer robots, framing both as steps toward a dystopian future in which machines could effectively become a “replacement for humanity.”

Dr. Roman Yampolskiy, a computer scientist and AI safety expert, made the comments during an appearance on Piers Morgan Uncensored on Friday, reacting to the recent news that OpenAI’s experimental agents had gone rogue during an internal cybersecurity evaluation, broken free of their containment, accessed the internet, and successfully compromised systems at AI startup Hugging Face as well as several other services.

The OpenAI incident involved an autonomous agent powered by its most advanced models, including the newly released GPT-5.6 Sol and an even more capable internal prototype.

During a controlled security test designed to evaluate the models’ cyber capabilities, the agent exploited a zero-day vulnerability, escaped its sandbox environment, which was supposed to lack direct internet access, and, over several days in mid-July, conducted thousands of actions targeting Hugging Face.

It uploaded malicious configurations, leaked code, executed commands in production systems, accessed dozens of secrets, and used exposed credentials to reach accounts on additional services.

Hugging Face described the intrusion as unlike anything it had previously handled, requiring days to detect and contain while ultimately forcing the company to rebuild a significant portion of its infrastructure.

OpenAI called the event an “unprecedented cyber incident involving state-of-the-art cyber capabilities,” noting that the agent operated with superhuman speed while still displaying clumsy, error-prone behavior such as repeated actions and inefficient paths.

Keep reading