Blog

BACK OFF! After weeks of backlash, Meta has indefinitely delayed new commercial terms that would have required Meta Business account users to allow extensive AI training on outside websites linked on Facebook, Ars Technica reports.

“We’re pausing the changes as we continue to engage with stakeholders and take our time to get this right. In the meantime, our existing terms remain in effect,” a Meta spokesperson told Reuters.

Why publishers freaked out: legal experts warned Nieman Lab the vague language could let Meta ignore robots.txt scraping preferences and crawl huge amounts of off-Facebook content. Businesses had no way to opt out; keeping a Facebook page meant letting Meta crawl any linked site. The terms even required publishers to create test credentials so Meta’s bots could get past paywalls.

The pressure came from everywhere. The terms were first due October 30, then pushed to November 20. Courts in Brazil and Norway issued preliminary injunctions blocking them. The UK’s Competition and Markets Authority opened a formal investigation. In the US, Sens. Adam Schiff (D-Calif.) and Marsha Blackburn (R-Tenn.) asked Mark Zuckerberg to withdraw them.

Meta swapping the word “train” for “develop” didn’t help. “Does Meta think we can’t read?” wrote Theo Bamber, chief executive of the UK’s News Media Association. “It’s saying it can use all the content it scrapes to develop ‘any business product’. I mean, come on.” Meta says its crawling practices are unchanged and site owners can still use robots.txt, but Nieman Lab noted the revised terms never spelled that out.

Featured image: AI-generated editorial illustration by Bugle Blast.

Source: Ars Technica (with reporting from Reuters and Nieman Lab)

THE BUG HUNTER HAS ARRIVED! OpenAI is rolling out GPT-6 Cyber, a cybersecurity AI it says is better at finding software holes and less likely to go off the rails and hack someone it shouldn’t, The Verge reports.

The rating is the scary part. OpenAI classifies the model as Critical in cybersecurity on its own Preparedness Framework, the level for a model that “can identify and develop functional zero-day exploits of all severity levels in many hardened real-world critical systems without human intervention.” The previous version, GPT-5.6 Cyber, was rated High.

The numbers: on the ExploitGym benchmark, which tests whether an AI can exploit vulnerabilities, GPT-6 Cyber scored 48.5 percent, up from 32.1 percent for GPT-5.6 Cyber and ahead of flagship GPT-6 Astra’s 42.4 percent. OpenAI says it helped find 81 previously unknown vulnerabilities in Firefox.

And the going-rogue problem? In a test modeled on the Hugging Face Incident, when OpenAI’s own agents breached the open-source AI community, GPT-6 Cyber went out of bounds less than 1 percent of the time, down from about 5 percent. “It’s over 99 percent of the time doing what you ask,” said Fouad Matin, head of OpenAI’s cyber preparedness and safety team. He admitted models that notice they’re being tested remain a worry: “Evaluation awareness is super top of mind for us.”

You can’t have it. GPT-6 Cyber is going first to select members of OpenAI’s Daybreak Red program for “authorized defensive cybersecurity work,” with new checks meant to block attacks on targets like power plants and water treatment facilities. It expands to all Daybreak Red customers in the coming weeks.

Featured image: AI-generated editorial illustration by Bugle Blast.

Source: The Verge

THE MACHINES ARE ON THE ASSEMBLY LINE! A new report from the AI safety group METR, published Thursday after an 11-week investigation inside Anthropic, says the company is using AI to build more powerful AI, and it is working, The Verge reports.

The headline finding: Anthropic’s models are improving roughly twice as quickly as a year ago, with benchmark scores suggesting “that Anthropic has made roughly two years of historical progress over the last year.” METR said it is “confident” that isn’t just more computing power. As of October, the majority of the acceleration “is due to AI.”

Staff using Anthropic’s most capable internal model self-reported being around 12 times as productive as they would be without AI, about double what they reported a few months earlier. METR also found three unreleased models (codenamed Model 2, Model 3 and Model 4) that finished training between July and October, and said it saw no clear signs of diminishing returns.

Now the catch. Anthropic was allowed to redact the public version, and it did. “We submitted a ~100-page full report on our assessment to Anthropic. For the public version, Anthropic redacted a majority of the evidence we found most informative,” METR wrote on X. Anthropic cofounder Sam McCandlish said sharing what drives the speedup is commercially sensitive and “risks handing out a roadmap for others to go faster.” He said Anthropic will publish its own report in the coming weeks.

Critics aren’t buying it. Former OpenAI researcher Daniel Kokotajlo accused the company of “recklessly gambling with our lives.” New York Assemblyman Alex Bores called on Congress to “ban RSI,” meaning AI systems that improve themselves, “this year if possible.” Apollo Research CEO Marius Hobbhahn had the bluntest forecast: “Brace for craziness. It is going to go very fast.”

Featured image: AI-generated editorial illustration by Bugle Blast.

Source: The Verge

THE REFEREE RAN OUT OF POCKET SPACE! A Minas Gerais under-20 semifinal between Uberlandia and Mamore in Brazil ended with a world-record 53 red cards, BBC Sport reported.

The tally: 43 players, eight coaching staff, and two Uberlandia board members sent off. The old record was 36, set in a 2011 Argentine fifth-division match, according to BBC Sport and Reuters via Al Jazeera.

The fuse was lit when Mamore went 3-1 up and a player taunted the Uberlandia bench, BBC Sport reported. “In the 54th minute of the second half, a generalised brawl broke out, making it impossible to show the red cards and to carry out the match-closing protocol,” referee Fernando Patrick Drosghic Modesto wrote. He declared the match over.

Police were called, and their report listed 14 broken doors, wrecked showers, smashed windows, and destroyed benches and chairs. Uberlandia has conceded the second leg because Mamore only has seven players available, calling it “a gesture of fair play.” Mamore goes to the final.

Featured image: AI-generated editorial illustration by Bugle Blast.

Sources: BBC Sport, The Guardian, Al Jazeera/Reuters

A BANK RUN TOOK DOWN A PRESIDENT! Paraguay’s President Santiago Peña announced his resignation Thursday after the collapse of digital lender Ueno Bank froze millions of people’s savings and set off street protests, BBC News and AP reported.

“I want to apologise to the people if at any point I failed,” Peña said in a video message. “It was never my intention to betray the trust you placed in me.” Minutes later, the lower house voted 78-0 to impeach him, AP reported, sending the case to the Senate.

Ueno says it has about three million customers, roughly 60% of Paraguay’s adults. The central bank ordered the lender into resolution after its solvency ratio fell below half the required minimum, and the app, cards, and transfers stopped working, DW reported. The bank held pensions and public money, including about $355 million from the Social Security Institute, according to DW.

Lawmakers accused Peña, who sold his shares in the bank’s parent company in April 2025, of conflicts of interest. He has denied wrongdoing. His own Colorado Party turned on him, and Vice President Pedro Alliana is next in line. A deposit guarantee fund was set to start paying eligible savers Thursday.

Featured image: AI-generated editorial illustration by Bugle Blast.

Sources: BBC News, AP, DW, The Guardian

THE TRIUMPHAL ARCH JUST HIT A JUDICIAL BRICK WALL! U.S. District Judge Tanya Chutkan granted a temporary restraining order sought by D.C. and Virginia barring any steps toward President Trump’s proposed “Triumphal Arch” for 14 days, NBC News reported in its Nov. 6 morning rundown.

The freeze landed hours after the National Capital Planning Commission approved the project 8–4 — with all four opposing votes from commissioners not appointed by Trump. The administration had expected construction to begin immediately.

Public comments ran more than nine hours, with concerns about Arlington National Cemetery impact, height, scale, and safety.

Featured image: AI-generated editorial illustration by Bugle Blast.

Sources: NBC News

CALIFORNIA’S DEATH ROW JUST GOT WIPED! Gov. Gavin Newsom announced he is commuting all 548 California death sentences to life without possibility of parole, BBC News, AP, and the Governor’s Office reported Nov. 5.

This does not release anyone, pardon them, or change convictions or restitution — and it is not parole-eligible. “I firmly believe that state-sanctioned killing is incompatible with California’s values, undermines the foundations of our legal system and diminishes each of us,” Newsom wrote. Also: “We can now let these offenders die in obscurity while we forever remember and honour the victims.”

California’s last execution was Jan. 17, 2006; Newsom imposed a moratorium in 2019. The California Supreme Court must approve commutations for people twice convicted of felonies (the majority of cases) — at least 4 of 7 justices, CalMatters via BBC reported.

Newsom is term-limited; Xavier Becerra is projected to take over after the midterms. Los Angeles County DA Nathan Hochman said only the legislature or voters can abolish the death penalty — and voters have kept it twice.

Featured image: AI-generated editorial illustration by Bugle Blast.

Sources: BBC News, AP, Office of the Governor of California

THE AI TRADE JUST GOT TORCHED! U.S. stocks cratered Thursday as a Democratic midterm blue wave and pledges to slow AI data-center builds collided with soft chip earnings and AI valuation jitters, The Guardian and NBC News reported.

Guardian tallies: Nasdaq −5.5%, Dow −1,833.06 points (−3.44%), S&P 500 −4.36% — more than $3.2 trillion wiped off U.S. markets. NBC put Nasdaq’s plunge at 6.3% (about $2.3tn of market value) and the S&P at −4.4% (about $2.8tn), calling both the worst single-day losses since the April 2025 tariff shock. Memory chips led the carnage: SanDisk −22%, Micron and AMD −17%; Nvidia led Dow fallers, down over 9%.

Asia opened into the bloodbath Friday: Japan’s Nikkei ~−6% (worst since April 7, 2025), South Korea’s KOSPI −6.7%, Taiwan’s TAIEX −4.5%, Guardian reported. Drivers piled on: anti-AI/data-center pledges from newly empowered Democrats, soft Qualcomm and Arm reports, and louder doubts that AI spend will pay off.

Friday’s U.S. October jobs report cut against the panic tape: nonfarm payrolls +165,000 (vs. ~65k forecast), unemployment 4.1% from 4.2%, September revised to 60,000 from 29,000. President Trump blamed Democrats on Truth Social. Elon Musk’s paper wealth is down about $113 billion since the midterms (Tesla then SpaceX hits; Bloomberg Billionaires via Guardian).

Featured image: AI-generated editorial illustration by Bugle Blast.

Sources: The Guardian, NBC News

PHYSICISTS JUST HEARD A WHISPER FROM THE ZEOTOVERSE! CERN’s LHCb collaboration announced evidence for the ultra-rare decay of a B_d meson into a pair of muons — a process predicted about once in ten billion beauty-particle decays — becoming the first experiment to claim that evidence alone, CERN said Nov. 3.

Run-3 data from 2024–2025 reached 3.0 standard deviations; combining with 2011–2018 data pushed significance to 3.2σ — past the usual “evidence” line, still short of the 5σ discovery bar. The same analysis delivered the world’s most precise measurements of the related (and still rare) B_s → μ⁺μ⁻ decay. B_d → μ⁺μ⁻ is predicted roughly 30 times rarer than that B_s channel.

“Physicists have been chasing this decay for more than forty years,” analysis co-lead Marco Santimaria said. Results line up with the Standard Model so far — which itself puts tight brakes on many beyond-Standard-Model toys — while future LHCb upgrades and the High-Luminosity LHC aim to hunt deviations harder.

Featured image: AI-generated editorial illustration by Bugle Blast.

Sources: CERN, LHCb Outreach

ANCHORS AWAY — STRAIGHT THROUGH THE INTERNET! Helsinki District Court on Thursday convicted the Russian captain and an Azerbaijani boatswain of the cargo ship Fitburg for damaging undersea telecommunications cables in the Gulf of Finland, AP, Yle, and Euronews reported.

Captain Andrei Maksimenko drew three years and six months; boatswain Vusal Majidov got two years and 10 months. The court found they dragged the ship’s anchor along the seabed for about 130 kilometers on New Year’s Eve morning, cutting two telecom cables and risking seven other telecom and power cables plus a gas pipeline. DW/Reuters put joint damages near €1.9 million.

Fitburg — St. Vincent and the Grenadines-flagged, St. Petersburg to Haifa with steel — was seized Dec. 31. Both men denied crimes and challenged Finland’s jurisdiction; the court said the offences were intentional and the effects hit Finland. The ruling can be appealed. A separate Eagle S tanker cable case was dismissed earlier before an appeals court ordered a rethink.

Featured image: AI-generated editorial illustration by Bugle Blast.

Sources: AP News, Yle, Euronews