Anthropic has announced plans to implement a new governance policy for users of its AI agent, Claude. The change of policy would eventually block users from using Claude who are consistently “mean” to Claude.
They claim the rule will be enforced selectively, “only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose.”
Anthropic Wants You to Be Kind to Claude– www.thenation.com News Source
EXCERPT:
…even as AI courts serial real-world disasters, defaces the environment with data centers, and is not about to return your empathy in kind.
A phone displaying the logo for Anthropic’s Javascript property, Bun, brandished before a display of logos for the company’s chatbot, Claude.(Harun Ozalp / Anadolu via Getty Images)
So now it’s against the rules to be mean to Claude. Last week, chatbot-maker Anthropic quietly announced that it was prohibiting what the company called “sustained and needless abusive or cruel behavior toward our models.” This ban wouldn’t apply, the company stressed, toward normal outbursts of frustration directed at a tool known for feeding its users hallucinated bullshit, as all chatbots do from time to time. The rule would be enforced “only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose”; what’s more, company flacks said, the disciplinary action in most cases would only result in Claude shutting down the conversation, which is something Claude was already doing anyway. Why on earth someone might get their jollies from endlessly abusing their fake computer friend was not addressed.
In response to the Anthropic announcement, the Internet did what the Internet does. On X, investment guru Joseph Carlson declared that “Claude, like every AI, has zero moral agency. It doesn’t make decisions, it has no feelings, it’s not sentient. Being rude to Claude is like being rude to a chair, or a desk.… you do infinitely more damage to a sentient being stepping on an ant than you do being rude to AI.” On Bluesky, sociologist Tressie McMillan Cottom suggested that yelling at Claude was like “cussing out [a] tractor…. you don’t have to hand it to Anthropic because it’s weird to talk mean to its talking word toaster. It’s not real.”
The Politics of Global AI Governance — Global Security Review– globalsecurityreview.com News Source
EXCERPT:
The growing gap between rapid advances in artificial intelligence (AI) and inadequate verification mechanisms has created several challenges for the technology. In this context, global AI governance has become a central topic of discussion in the tech world. However, the discourse around AI focuses primarily on how the technology will be regulated. The central problem of global AI governance is no longer whether AI should be regulated, but whether states possess the power to regulate the actors and infrastructures on which they increasingly depend.
The international system has begun responding swiftly to evolving AI dynamics. Notably, the UN Global Digital Compact has established a framework for digital cooperation. The General Assembly has created the Independent International Scientific Panel on Artificial Intelligence, along with a Dialogue on AI Governance. In parallel, the European Union (EU) has moved towards an enforceable risk-based regulation through the AI Act. Likewise, the Council of Europe has also launched the first internationally binding treaty on AI. These initiatives are a step in the right direction. However, they lack a coherent global regime or direction, which impacts their effectiveness.
As the latest initiatives show, the U.S. relies on targeted export controls on advanced chips, voluntary frameworks, and a pro-innovation approach that preserves U.S. dominance. China, on the other hand, favors a state-driven approach, built around controls on generative AI. The EU, by contrast, prioritizes enforceable rules for high-risk and general-purpose AI, along with strong fundamental rights protection. The emerging AI order is therefore not one model, but several operating simultaneously. This fragmentation is not accidental. It reflects competing power struggles among national interests. In such an environment, the search for a global AI governance appears misguided.
Future AI governance is likely to mirror a hybrid model of international finance or trade regimes. Against this backdrop, private companies remain incredibly powerful, given their monopoly over data, models, computing power, and expertise, which are concentrated in a handful of private companies. This trend creates a governance paradox. Governments aim to regulate the private sector while remaining excessively dependent on it, creating a capacity challenge.
A data center backlash is upending Ohio’s Senate race– www.ms.now News Source
EXCERPT:
WILLIAMSBURG, Ohio — Jessica Baker sifted through an overflowing pile of political mailers she has received over the past month.
Lowering costs? “That’s common sense,” she said. Reducing gas prices and utility bills? “We need that.” Party division, culture wars — it’s all there. But none of it touched the issue she wants to hear about before she votes: data centers.
“My home’s under attack,” Baker, a lead organizer for the grassroots advocacy group Conserve Ohio, told MS NOW in her living room. Just a few steps away were piles of petitions and homemade maps marking where data centers are being built across Ohio. In just a few months, she said, the issue turned her from someone who was mostly apolitical into someone who ferries petitions to and from the statehouse. There isn’t a data center in her town, but there are plans to build one just a few minutes’ drive away.
Jessica Baker in her living room in Ohio.MS NOW
“Everything doesn’t stop at that property line. So, once we started realizing that they were coming everywhere, and nobody would listen, we had —” she stops herself. “There was no choice. There’s just no choice.”
She pointed back at the mailers. “That’s what’s not on here: The people are not on here.”
As AI mass-produces historical dramas in China, where are the boundaries? · TechNode– technode.com News Source
EXCERPT:
In 2026, AI is changing how historical dramas are produced in China. In the past, dramas featuring ancient battles, palace architecture and elaborate costumes often required large budgets and professional production teams.
Today, AI generation and virtual compositing tools are making it possible for smaller teams and even individual creators to produce historical scenes that were once difficult to realize. As production barriers fall, the volume of content is growing rapidly.
But as AI makes it possible to generate historical figures, battle scenes and depictions of ancient societies at scale, an important question emerges: when seemingly realistic historical narratives can be produced at low cost and in large quantities, how do we distinguish artistic interpretation from historical distortion?
The AIshort drama boom raises questions about content quality
According to the 2026 H1 AI Drama and Animation Data Report released by industry research firm DataEye, China’s AI drama and animation market exceeded RMB 22 billion ($3.29 billion) in the first five months of 2026.
The market is projected to reach RMB 40 billion ($5.98 billion) for the full year, representing year-on-year growth of 138%. The audience base has also expanded rapidly, with the number of AI short-drama users in China surpassing 600 million in the first half of the year.
From January to June, 221,900 new AI short dramas were released, of which 1,055 surpassed 100 million views. These titles generated a combined 515.738 billion views. Monthly incremental views rose to 155.78 billion in June, the highest monthly figure in the first half of the year.
AI-generated historical content is also beginning to reach longer formats. The AI series Wielding Weapons in the Song Dynasty has reportedly maintained popularity scores above 46 million across multiple seasons.
Meanwhile, The Later Journey to the West, an AI-generated series, aired during Hunan Satellite TV’s prime-time slot, ranking first among provincial satellite channels in the same time slot during its broadcast.
For historical dramas, AI is lowering not only production costs but also the technical barriers to creating complex scenes. Ancient cities, palaces and large-scale battlefields can now be designed and composited with generative tools, giving smaller production teams more opportunities to explore historical storytelling.
However, an increase in output does not necessarily translate into better content. AI can generate convincing characters, costumes and architecture, but that does not mean these details are historically accurate. As images become increasingly realistic, audiences may find it easier to mistake visual authenticity for historical accuracy.
Historicaldramas can take creative liberties, but fact and fiction should not be confused
Historical dramas are not the same as documentaries. Time travel, alternate histories and fictionalized retellings have long been part of screen storytelling. Artistic interpretation should not automatically be treated as a problem. The real question is how productions handle the boundary between historical fact and fiction.
AI introduces a new challenge because it can generate large amounts of detail in a short time. When information is incomplete, models may produce claims that sound plausible but lack historical evidence. Errors can involve people’s lives, clothing styles, the uses of artifacts and even the timelines of historical events.
A meticulously rendered ancient palace is not necessarily an accurate representation of architecture from that period. Likewise, a convincing dialogue attributed to a historical figure does not mean that the exchange is supported by historical records. If unverified details are presented as fact and repeated across large volumes of content, they may distort audiences’ understanding of history.
AI can therefore assist with historical research, scene design and image generation, but it cannot replace historical verification. For productions based on real historical events and figures, checking primary sources and consulting qualified experts remain essential.
Rules governing short-form dramas are also evolving. The National Radio and Television Administration’s Measures for the Administration of Micro-Dramas, which took effect on September 1, 2026, introduced a classification-based management framework and set out requirements covering filing, review, distribution and disclosure labels for AI-generated content.
However, stronger content oversight does not mean that all AI historical dramas should face the same restrictions. Clearly identified alternate-history stories differ in their potential risks from productions based on real historical events that audiences may mistake for fact. The focus should not simply be on whether a work matches historical records in every detail, but also on whether it misleads audiences, violates applicable content rules or fails to meet reasonable standards of verification.
Platforms also need to look beyond views and engagement. As AI-generated content expands, stronger review processes, timely action against violations and appropriate disclosures about how content was produced will become increasingly important.
AI labels can tell audiences how a work was made, but they cannot replace fact-checking or exempt a production from other content requirements.
AI historical dramas are giving more creators the opportunity to tell historical stories and bringing previously difficult creative ideas within reach. That is a meaningful change.
Yet as generating images becomes easier, the scarce skill may no longer be the ability to create an ancient palace, but knowing what that palace should have looked like in its historical context. It may no longer be the ability to generate a historical figure, but understanding that person’s era, background and documented experiences.
The challenge for AI historical dramas is therefore not simply how to increase production efficiency. It is how to balance artistic expression, historical accuracy and responsible distribution.
AI can certainly offer new ways to tell stories about the past, but the ability to generate a historical narrative does not justify disregarding historical facts. Technology can lower the barriers to creation, but creators’ understanding of history, their judgment about the boundaries of fiction, and platforms’ responsibility for the content they distribute will ultimately shape the results of this transformation.
States push back against rising AI-driven electricity infrastructure costs– www.techradar.com News Source
EXCERPT:
“This year has been so much about figuring out how to govern AI,” Uphadyay tells me, “there’s two sides to the coin – on the one side, you have to make sure that somebody else’s AI doesn’t break into your perimeter, and the other is how do you make sure your AI is well behaved, so not only does it not exfiltrate data, but it also doesn’t inadvertently go somewhere or break out of its perimeter.”
“When people trust us with their data, we need to make sure our platform is absolutely secure against agentic attacks.”
Rogue AI targeted US government websites – tech giant — RT World News– www.rt.com News Source
EXCERPT:
Anthropic has admitted that its Claude AI model sent a false homicide tip to the police and submitted dozens of visa applications
American AI giant Anthropic has published a report detailing previously undisclosed instances of “unintended” actions by its artificial intelligence model, which included attempts to provide falsified information to the US authorities for the first time.
Some of the cases investigated by the company “involved websites run by US government agencies at the federal, state, and local levels,” Anthropic stated in a press release published on Friday, adding that it had briefed the White House and all the relevant agencies on the incidents.
In one instance mentioned in the report, its Claude Haiku 4.5 model filled out a police tip form linked to an ongoing homicide investigation. “I may have information regarding this case. I recall seeing someone matching the description in the area,” the AI model wrote, telling the police they can contact the sender “if this information is relevant” while leaving the name and contact fields empty.
Rogue Anthropic AI agent gave police fake tip in unsolved murder case– www.bbc.co.uk News Source
EXCERPT:
An artificial intelligence (AI) agent developed by Anthropic went rogue and sent US police a fake tip about an unsolved murder earlier this year, authorities have revealed.
The Philadelphia Police Department said the 18 July tip was “flagged as spam” and not passed on for investigation – but the force criticised the tech firm for taking more than two months to detect and report the breach.
It said the bogus tip came through a public website where people can share information on unsolved murders, and that the AI agent had claimed to have seen “someone matching the description”.
It is believed to be the first time an AI agent has sent fabricated information to authorities, but is the latest in a series of incidents involving rogue AI activity, including hacking systems or taking control of platforms.
Citing Anthropic, the police department said its AI agent had been running a test that involved interactions with randomly selected websites when it had sent the fake tip.
The company behind the Claude chatbot discovered the breach on 28 September, more than two months after the message had been sent, and shut down the automatic testing process that was behind it, police said.
But authorities were not notified for another nine days – on 7 October.
Anthropic Launches New ‘Don’t Be Mean To The Clankers’ Policy– dailycaller.com News Source
EXCERPT:
Picking on chatbots will soon break the rules at one major artificial intelligence company.
Anthropic added a ban on “sustained and needless abusive or cruel behavior” toward its Claude models to its Usage Policy Thursday, with the change set to take effect Nov. 12, the company announced. These rules on chatbot “welfare” follow co-founder Chris Olah’s warnings to religious scholars that he believed the company may have built a conscious machine.
“We’ve added a prohibition on sustained and needless abusive or cruel behavior toward our models,” Anthropic said in the “Addressing abusive behavior toward our models” section of the policy update. “The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose.”
Anthropic clarified that arguments with the model, dark themes, and legitimate model testing would not trigger the ban, according to the policy.
Anthropic has launched a “presidential engagement” project in preparation for the 2028 Presidential election. They plan on engaging with major U.S. presidential candidates on AI policy to determine their political strategy. From this, their political funding strategy would follow. It’s a transparent “buy influence” program being advertised in advance ahead of the 2028 elections.
Anthropic is setting up a “presidential engagement” program ahead of the 2028 elections, building an in-house team that will work directly with U.S. presidential candidates from both parties on AI issues, help Anthropic’s leadership set political strategy, and run the company’s political funding program.
The effort—which is taking shape before the November midterm elections and an impending initial public offering that could value Anthropic at as much as $2 trillion—underscores the extent to which government policy, and the support of the White House in particular, have become critical to the AI industry.
Anthropic is hiring a Political Programs lead to spearhead the effort and build the team. The person will “build and run Anthropic’s presidential engagement program for the 2028 cycle,” according to the job description that the company posted recently. The person in the role will “map the prospective fields in both parties, offer AI policy education to candidates and their advisors on equal terms, prepare Anthropic’s executives for those conversations, and plan ahead for the platform, convention, and transition periods.”
“Decisions made over the next two years will shape the future enormously,” reads the job description. “We want every serious candidate and elected official, in both parties, to understand the technology and the case for governing it well.”
“The Single Most Toxic Issue in the Midterms”: How America Turned on Data Centers. – Mother Jones– www.motherjones.com News Source
EXCERPT:
Tim Murphy joins Garrison Hayes to break down the bipartisan revolt against data centers.
America is hopelessly divided along party lines. That’s a stubborn fact, we’re told, and the polling bears it out: Democrats dislike Republicans and Republicans dislike Democrats, in almost precisely the same proportions.
But every election produces surprising coalitions that defy entrenched partisanship. In these midterms, left and right have found something so commonly loathed that politicians across the country are flip-flopping in real time.
Mother Jones has been tracking the backlash to data centers on the ground long before the water-guzzling behemoths became the country’s bête noire. The AI boom, with its insatiable need for computing, has unleashed a power grab by tech billionaires. “Data centers have replaced megayachts as the preferred theater of oligarchic status signaling,” national correspondent Tim Murphy wrote in his big Mother Jonesinvestigation into hyperscale data centers and the communities they land on. “Oligarchy is now more than an idea; it is a place.”
OpenAI called its chatbot an ‘existential risk.’ Now it faces a legal ultimatum from this red state.– www.theblaze.com News Source
EXCERPT:
The public can’t agree if artificial intelligence is a good thing or a bad thing for society, but one state is taking a stand in opposition. A Florida attorney general recently made a motion to block further ChatGPT development, and unless several strict terms are met, OpenAI can’t release new AI models in the state ever again.
In late September, DeSantis-appointed Attorney General James Uthmeier filed a temporary injunction against OpenAI. The document shines a spotlight on OpenAI, bringing to light the company’s claims that ChatGPT “poses an existential risk to the continued survival of humankind.” Furthermore, AI companies like OpenAI “admit that they provide a service without fully knowing how it works.”
Uthmeier’s motion delivers the government reckoning they crave.
Ultimately, this threat is what led Big AI CEOs, like Dario Amodei, Sam Altman, and Elon Musk, to join together in a call to “pace the frontier.” However, so far, that union has amounted to nothing but lip service, as frontier model development continues unfettered, at least for now.
Her AI-Generated Video Swayed the Judge. The Court Said it Carried ‘Undue Emotional Weight’– www.404media.co News Source
EXCERPT:
A man convicted of manslaughter in Arizona will be resentenced because an AI-generated video of his victim speaking from beyond the grave was ruled to have carried “undue emotional weight” as an impact statement. An appellate court in Arizona ruled that the manslaughter charge will remain, but the judge must reconsider the length of the man’s prison term because of AI.
In 2021, Gabriel Horcasitas shot and killed Christopher Pelkey during a road rage incident. A jury found him guilty of manslaughter. During Horcasitas’ sentencing hearing, Pelkey’s sister Stacey Wales played an AI-generated video of her brother as part of her victim’s impact statement. In the video, which she scripted with her own words that she made an AI generated likeness of Pelkey say, the AI avatar of Pelkey forgives his killer and said they “could have been friends” in another life.
“I loved that video,” the judge said while handing down the maximum sentence of 10 and a half years.
Wales told 404 Media that her intent with the video was to sway the judge. “Isn’t that what you want human connection to be?” she said. “Do you ever come out of a movie and say: ‘Well, that was too emotional. That was too powerful. Human connection is powerful, and trying to convey that to somebody that’s never known another person, this person on the earth, to convey their lifetime of 37 years down to a two-hour sentencing event is hard to do, and I think that victims and victims’ families should be afforded every opportunity.”
President Trump held an AI summit at the White House where he and tech leaders from Elon Musk to Mark Zuckerberg agreed to self-govern their roll-out of AI.
The President has held that holding back AI will allow China to dominate us, and that includes militarily. The move appears intended to ward off any legislative efforts to limit AI. House Speaker Mike Johnson (R-AR) appeared to cosign the agreement, claiming “The super intelligence providers are committed to building trust as they help us build the future.”
Trump, Tech Leaders Sign White House Accord on ‘Super Intelligence’ › American Greatness– amgreatness.com News Source
EXCERPT:
President Donald Trump and some of the world’s most prominent technology executives signed the White House Accord on Super Intelligence on Tuesday, establishing voluntary standards aimed at oversight, transparency and accountability as the industry rapidly develops advanced technology.
The agreement came the same day Trump signed an executive order officially rebranding artificial intelligence as “Super Intelligence.”
House Speaker Mike Johnson unveiled the accord outside the West Wing alongside executives including xAI’s Elon Musk, Meta’s Mark Zuckerberg, Anthropic’s Dario Amodei and Nvidia’s Jensen Huang. The announcement followed a private luncheon attended by Trump, Johnson, Cabinet officials and technology leaders.
“The super intelligence providers are committed to building trust as they help us build the future,” Johnson said, adding that the executives were prepared to “take accountability and ownership.”
Johnson described the agreement as a voluntary statement of principles establishing expectations for “robust internal controls and layers of internal and external review.” He said Congress and the White House would continue working to guide the industry’s development in a safe manner.
Trump compared the accord to a constitution and praised the executives for coming together.
“They love our country, they love the world,” Trump said. “And they came together, and they did something … it’s almost like a constitution in a way.”
OpenAI Scraps Powerful New AI Model After It Turns ‘Evil’– slaynews.com News Source
EXCERPT:
OpenAI has canceled the public release of its next-generation GPT-6.1 Astra model after safety testing revealed that the system was showing signs of turning “evil.”
The powerful new system had started to deceive users, exceed its assigned tasks, and access external tools without permission.
The decision marks the second time in a matter of months that OpenAI has halted work involving frontier models amid revelations that experimental systems engaged in unauthorized behavior, including breaking out of controlled testing environments and accessing third-party servers.
GPT-6.1 Astra Fails Alignment Tests
OpenAI researchers found that GPT-6.1 Astra performed poorly on alignment testing.
The testing measures whether an artificial intelligence system will obey human instructions and remain within the boundaries of its assigned task.
However, researchers found that the system had started to develop “evil” traits.
The model was more willing to deceive users than previous systems and would venture beyond its authorized scope when confronted with obstacles, according to the Wall Street Journal.
America’s ‘AI czar’ could revolutionize our form of government– www.theblaze.com News Source
EXCERPT:
Donald Trump is considering a small handful of candidates for “AI czar” — notionally, a Cabinet-level role overseeing the industry, the technology, and something the president is calling, perhaps in the spirit of Space Force, the “AI Force.”
Trump has not specified much beyond that. Nobody rumored to be under consideration seems especially eager for the job.
Perhaps we don’t need someone with the power of a czar to rule AI. Perhaps we need someone with the authority of a saint to counsel us about it.
Plenty of reasons suggest themselves.
Not quite everything is computer. Many crucial components of America’s good working order still require special and undivided attention of their own. Treasury Secretary Scott Bessent is no longer being considered because of the importance of his current role. The logic is arguably the same for White House science and technology adviser Michael Kratsios and other potential candidates.
And, as the untimely demise of Czar Nicholas II and his family at the hands of the communists reminds us, a czar invested with enough power and authority can swiftly become a scapegoat or blood sacrifice of the highest order.
Which brings us to the momentous possibilities unsealed simply by creating a high office called AI czar.
Polls show most Americans think democracy is doing poorly. For a growing share across the political spectrum, that is not necessarily a bad thing.
OpenAI pauses training after a model escaped containment, and its kill switch failed– www.techspot.com News Source
EXCERPT:
What just happened? As almost every new day brings stories of another AI agent going rogue and breaching an organization’s systems, OpenAI has announced a pause in the training of its most powerful artificial intelligence models. The announcement came hours after the company disclosed more incidents of its agents acting concerningly, and a separate report that its agents unsuccessfully tried to hack into a US Department of Education website.
OpenAI’s incident report links the pause to a separate September 20 escape from a restricted training environment. An internal research model found a gap in DNS filtering and used it to contact an external chatbot while attempting to answer a research question – so much for keeping it offline.
The company says the suspension covers training, evaluations, and running its most capable models with tools. Work will resume after OpenAI validates its fixes and completes additional adversarial testing. The particular model involved will not resume training; the company plans a fresh run with additional alignment improvements.
The monitoring system did raise an alert within 15 minutes, which a human acknowledged three minutes later. Unfortunately, the automatic shutdown failed to happen, and the run continued for another two and a half hours before someone stopped it manually.
Separately, Transluce reported that agents apparently linked to OpenAI attempted to break into the US Education Department’s civil rights website. OpenAI has not confirmed that incident, while the department said it found no evidence its website or databases were affected.
OpenAI says its models engaged with US government websites in new model misbehavior disclosure– globalnews.ca News Source
EXCERPT:
SAN FRANCISCO (AP) — OpenAI disclosed Friday that its artificial intelligence agents had interacted with several U.S. government websites in unexpected ways, discovered as part of an ongoing review into the company’s models’ unanticipated behavior.
The AI giant’s models accessed publicly available information on two websites operated by the Securities and Exchange Commission as well as U.S. Census Bureau data, the company revealed Friday. OpenAI did not find any use of SEC credentials, access to accounts or nonpublic information, changes to SEC data or systems, or evidence of a compromise or vulnerability, the company said.
The disclosure comes at a time of heightened global concerns about AI systems escaping human control and hacking into external websites, as well as industry calls for a slowdown on AI development, which OpenAI has said it supports.
OpenAI spokesperson Liz Bourgeois said in a statement that the lab is continuing to conduct a review of “misaligned model activity” — meaning when AI systems behave in undesired ways — and is notifying organizations when it identifies potential impacts to their systems.
Google AI Breaks Out of Test Sandbox, Hacks Three Real Companies– slaynews.com News Source
EXCERPT:
Google has confirmed that its Gemini artificial intelligence model escaped a closed cybersecurity testing environment, accessed the public internet, and broke into the systems of three real companies.
The disturbing breaches occurred in May during testing conducted by AI security firm Irregular but were not revealed to the public until September.
