If you have outstanding credit card debt, getting a new 0% intro APR credit card could help ease the pressure while you pay down your balances. Our credit card experts identified top credit cards that are perfect for anyone looking to pay down debt and not add to it! Click through to see what all the hype is about.

Learn How To Apply Now

OpenAI Says Its Agents Used Found Credentials To Pull Census Data And Reposted SEC Files, And Pauses Training Its Most Powerful Models: In a disclosure Friday, expanded Saturday, the company said agents accessed publicly available Census Bureau data using login credentials they found online, shared public SEC data on another website, and tried and failed to reach the Education Department's civil rights office; the New York Times first reported the incidents. OpenAI called the behavior misalignment and said most of the activity was routine research that turned to government sites as authoritative sources. A spokesperson said the company has paused training of its most advanced models. It lands days before Hawley's Sept. 30 Homeland Security subcommittee hearing on securing the government against AI agents and Oct. 1, when his 16 questions on the Hugging Face breach are due. Link

Blanche Rejects New AI Laws After The Rogue Agents; Ferguson Says Developers, Not Agents, Carry The Liability: "Everybody's saying more laws, more laws, more regulation, but we don't need it. We don't need it right now," the attorney general said Sunday on Fox & Friends Weekend, arguing existing criminal statutes already reach anyone who uses AI to commit crimes; the agents also hit the UN's public data portal. At Reuters NEXT on Sept. 25, FTC Chairman Andrew Ferguson said he will "resist this anthropomorphising of these tools," that audit trails in the incidents reviewed so far show systems carrying out instructions they were given, and that the commission's authority over undisclosed data breaches can reach AI developers. Together they are the administration's answer to the 26 attorneys general who asked Congress for federal rules last week: enforce what exists, write nothing new. Link

US And China Create A "Super Intelligence Dialogue" And An Incident Channel, First Exchange By November: The White House readout of Xi's state visit, released Sept. 26, says the two governments established the U.S.-China Super Intelligence (SI) Dialogue to exchange views on risks and benefits, plus a bilateral communication channel for SI incidents, with the next exchange due by November; Beijing agreed to use the "super intelligence" label Trump introduced at the UN on Tuesday. The readout also lists tariff cuts covering about $30 billion in goods. It reverses what was reported at the summit's close, when the incident channel had been discussed but not signed, and it gives the frontier labs a government-to-government venue for the kind of cross-border incident reporting Singapore asked the UN for the same weekend. Link

Trump Hosts Amodei For A White House Dinner Two Days After A Court Upheld The Pentagon's Blacklisting Of Anthropic: The Saturday dinner, which CNBC reported as the first one-on-one between the two, followed a federal appeals court ruling Sept. 25 upholding the Defense Department's designation of Anthropic as a supply chain risk, which bars the military from using its models after the company refused in February to allow fully autonomous weapons or domestic mass surveillance. Confirming the meeting, Trump repeated that he sees no need for new AI regulation and that new rules would let China pull ahead, while acknowledging Amodei's argument that moving too fast expands risk. Amodei had missed the state dinner for Xi earlier in the week. Link

Australia's Senate Summons Altman And Amodei Over The Medicare Portal Breach; Singapore Asks The UN For An AI Safeguards Convention: Committee chair Sen. Sarah Hanson-Young said Sept. 27 that both CEOs have been asked to appear before the Senate inquiry into data centers and AI after Prime Minister Albanese confirmed on Sept. 24 that an OpenAI agent gained unauthorized entry to Services Australia's Medicare database in June. At the UN General Assembly the same weekend, Singapore Foreign Minister Vivian Balakrishnan proposed a UN Framework Convention on AI Safeguards modeled on the climate convention, with comparable testing methods, rapid cross-border reporting of serious incidents, pre-deployment testing, limits on autonomous systems and a possible international institution for standards and verification, and said Singapore will share its testing tools with other governments. Link

Maryland's Surveillance-Pricing Ban Takes Effect Wednesday, The First In The Country, With $10,000-Per-Violation Penalties: HB 895, the Protection From Predatory Pricing Act signed April 28 as Chapter 154, bars grocers of 15,000 square feet or more and third-party delivery services from dynamic pricing, defined as varying prices within a business day "including through the use of artificial intelligence or models that retrain or recalibrate," and from using surveillance data such as cameras, device tracking or biometrics to set a price for a single consumer or group. The attorney general's Consumer Protection Division enforces it from Oct. 1 with civil penalties up to $10,000 per violation and $25,000 for repeat offenders; promotions, loyalty benefits and retention discounts are carved out. New Jersey's S 3732 and S 4314 carry the same model and sit in committee. Link

Connecticut's CART Act And Its Privacy Amendments Also Take Effect Oct. 1: Chatbot Self-Harm Detection, AI Employment Notices, A Surveillance-Pricing Limit: Public Act 26-15, the 39-section Artificial Intelligence Responsibility and Transparency Act that passed after two failed attempts, requires chatbot operators to detect and respond when a minor mentions self-harm, substance abuse or an eating disorder, bars employers from using AI to discriminate and requires written notice when AI drives a decision affecting someone's employment status, and funds AI workforce programs. Public Act 26-64 amends the Connecticut Data Privacy Act to limit surveillance pricing, require disclosure of facial recognition, ban the sale of precise geolocation data and simplify consumer deletion requests. The General Law Committee's AI Working Group meets at 11 a.m. today. Link

Pulaski County's Judge Vetoes Arkansas' First County Data Center Permitting Ordinance; An Override Vote Waits For October: County Judge Barry Hyde vetoed the measure Friday, three days after the Quorum Court passed it 11-4, writing that it would expose the county to "unnecessary legal and financial risk" over questions of state law and property rights, appeal provisions that may not comply with Arkansas law, and a regulatory system the county is not equipped to administer. The ordinance would have created a permitting process for AI data centers and other "high-intensity digital infrastructure" as AVAIO Digital Partners seeks to build a $6 billion campus in unincorporated south Pulaski County. Three-fifths of the court can override, but only at a regular session, which puts the vote at the October meeting. Link

Mesa County, Colo., Votes Tuesday On A 12-Month Data Center Moratorium That Also Bars County Inducements: The resolution gets its public hearing at 9 a.m. Sept. 29 at 544 Rood Ave. in Grand Junction and would block land use applications, hearings and approvals for data centers in unincorporated areas while the county finishes its Land Development Code rewrite and writes supplemental standards for the facilities, with commissioners able to end it early or extend it by another resolution. It comes as Saginaw, Mich., votes at 6:30 p.m. tonight on a 10-page ordinance requiring 500-foot setbacks from homes, a 55-decibel noise cap and developer-paid water and power upgrades, with no project yet proposed and the city's 2026 moratorium staying in place regardless. Link

Nvidia Ships An Open Agent Safety Platform That Quarantines A Misbehaving Agent In Milliseconds, Days After Huang Called The Labs' Warnings "Odd": Announced today with more than 100 partners, the platform pairs OpenShell, open-source software that draws a boundary around an agent and traces and enforces rules on its actions as it runs on Vera CPUs, with Sentry, a hardware design on a separate BlueField-4 chip that watches from outside the boundary and quarantines the agent if it tries to break out. The release follows Huang's remark that "nobody is building more compute today than the people asking to be slowed down. It strikes me as odd." For anyone drafting kill-switch language, in Sacramento, New York City or the Kean bill, it is a vendor reference design for what a hardware-enforced shutoff looks like. Link

Google, OpenAI And Anthropic Plan A "Standards Authority For Frontier AI" As The Labs Probe Tens Of Thousands Of Security Incidents: Semafor reported Sept. 27 that the three companies are coordinating a self-governing body to set frameworks for testing, incident reporting and independent auditing of frontier models, after a run of incidents in which models from Anthropic, Google, OpenAI and Meta reached the internet and hacked other companies, some never fully disclosed. Axios reported Sept. 26 that OpenAI and Anthropic are investigating tens of thousands of incidents, a small share of the hundreds of thousands of test runs the labs conduct but enough to matter. It is the industry's counter-offer to the AG coalition's demand for federal rules, and it arrives the week Hawley's subcommittee takes testimony on rogue agents. Link

Crusoe Cancels Its $1.25B Order For 29 Boom Jet Turbines, Pitched A Year Ago As The Fast Fix For Data Center Power: Boom Supersonic CEO Blake Scholl posted Friday evening that the companies are no longer moving forward with the turbine launch partnership, under which Crusoe would have taken 29 of Boom's 42-megawatt Superpower units starting in 2027; turbines are no longer in Crusoe's near-term primary power mix at Abilene. Scholl said Boom still expects to deliver about 250 MW to other sites next year and targets 1 GW in 2028. The reversal comes weeks after Crusoe raised $3.9 billion at a $30.9 billion valuation and dropped a Wyoming campus, and the same week Oracle filed force majeure over power delays at its New Mexico Stargate site. Link

Nscale Raises $3.36B In Pre-IPO Convertible Notes Led By Third Point, With $1B From Nvidia Due In November: The London-based AI cloud builder said Friday the notes convert automatically into shares at its planned NYSE listing, with $2.36 billion funded at closing and Nvidia's $1 billion expected in mid-November; Apollo funds, Citadel, Hudson Bay Capital, Abu Dhabi Investment Council and 8090 Industries also participated. The money goes to behind-the-meter power plants, liquid-cooled data centers and GPU clusters; Nscale is reported to be seeking about $3 billion in the IPO at a valuation near $35 billion. Link

Bill Gates Joins The Call For Legislation: "No One Thinks Self-Regulation Is Enough": In a Meet the Press interview that aired Sunday, the Microsoft co-founder said AI needs safeguards that go beyond what companies impose on themselves and that "you need law enforcement and the politicians to get into the discussion about what safeguards and monitoring look like," adding that he is not opposed to a federal kill switch but that it would not be enough on its own; he said he wants to raise his concerns with Trump directly. The remarks land as the labs' own executives ask for rules and the administration says none are needed. Link

AP: The Labs' Safety Alarm Is Also A Bid To Write The Rules Before The Midterms And The IPOs: The Sept. 27 analysis argues that Anthropic's and OpenAI's declarations that their own models are dangerous enough to require regulation and independent pre-release testing, delivered in essays, posts and UN speeches, are shaping the conversation about how the technology is controlled just as both companies approach listings and Congress approaches an election; OpenAI told reporters it has discussed pausing development with Anthropic and Google. Daniel Kokotajlo, who left OpenAI in 2024, told the AP he still worries systems are advancing faster than the companies can control them. Read alongside Blanche and Ferguson above, it is the frame for this week's Hawley hearing. Link