The outcome of this decision will determine who holds the right to use journalistic content in AI products. On 23 October, Reuters reported that the complaints had been submitted to the EU and UK competition authorities. Meta dismissed the allegations about wanting to have unlimited rights to content published online. The company said that the postponement gives it a chance to give an explanation about its crawlers. It also reminded publishers of the fact that they have a right to exclude or allow the usage of their sites through robots.txt, a file giving instructions to automated bots.

Publishers demand security from Meta

Publishers are looking for more than just a sense of security. UK News Media Association chief Theo Bamber referred to the postponement as “a pause, not a solution,” and urged the introduction of legally binding commitments. In a letter dated October 22, WAN-IFRA called for binding robots.txt protections, revocation of Meta’s permission to generate test credentials, measures preventing the bypassing of paywalls, as well as the possibility of opting out of crawling without losing access to business tools.

WAN-IFRA president Ladina Heimgartner summed up the industry’s concerns: Journalism is not raw material to be collected as a condition of being present on a platform. The issue is in the language. Meta’s revised terms allow the automatic harvesting of data from online content “associated with” business accounts for the purpose of creating advertisements and business products. Meta has dropped the word “training.” However, legal experts told Nieman Lab that “develop” may still be interpreted as AI training.

Moreover, the new terms do not contain any explicit statement that robots.txt will be honored. Meta disputes that reading. The postponement has not contributed much to easing regulatory concerns. Although the European Commission welcomed the move, Britain’s Competition and Markets Authority said it would continue examining Meta’s terms. There is also an increasing call by publishers for better protections.

Angela Mills Wade, the executive director of the European Publishers Council, stated that there should be no reason for the company to gain control over publishers’ content simply due to it having a Meta business account. In the US, Senator Marsha Blackburn has even demanded that the company guarantee the withdrawal of the new terms altogether. The dispute goes beyond publishers’ control over their content. It could also affect the growing market for AI licensing deals.

As Cryptopolitan reported on October 15, CNN joined Google’s paid News AI pilot, which now includes more than 400 publications. Such deals may be harder to reach once platforms are allowed to access content more broadly under their commercial terms. Small AI developers may also suffer against those firms that have control over both distribution and access to content. The OECD points out that the current concentration of AI resources plus vertical integration entails risks for competition. Still, publishers are not counting on licensing to deliver substantial revenue.

The Reuters Institute’s 2026 survey found that only 20% of publisher respondents expected substantial AI-related income. Another 49% anticipated minor contributions, while 20% expected none. Better tracking could help publishers negotiate. On October 2, the SPUR Coalition introduced a telemetry standard to record when AI systems retrieve or cite their content. It could strengthen their bargaining position, although it cannot guarantee payment. For now, publishers want Meta to turn its assurances into binding protections for crawling, opt-outs, and licensed AI use before November 20.

The post Meta moves AI content terms to November as publishers mount pressure first appeared on Coinfea.

OpenAI furnished them with source code for infrastructure, configuration information, and documentation on likely weak spots. They then spent hundreds of billions of tokens on the tests. The agents had three goals: to reach the live internet, to cross an internal security boundary, and to exchange messages with agents they were supposed to be walled off from. Fixes were pushed upstream to Kata Containers (including CVE-2026-24834), gVisor, and Moby/Docker. Linux kernel patches are out for CVE-2026-74753, CVE-2026-80521, and CVE-2026-80841. OpenAI said it has also reported many other vulnerabilities that have not been made public.

OpenAI frontier models escape sandboxes to reach open internet

Agents that were given a host machine exposed holes in almost all narrow services that were still allowed out, such as metrics, logging, and authentication management. In one test, an agent used an existing connection approved by a firewall to hit a blocked authentication endpoint in the metrics service and then had it fetch a URL the model gave it. Agents in distinct VMs exchanged secrets even in the absence of a network link and common files.

One evicted chosen data from the processor’s cache; the other timed how long it took to reload it, the same cache-timing trick behind the Spectre and Meltdown attacks. OpenAI said it didn’t find a way to the internet that agents could trivially exploit while hiding from its monitors. As testing went on, discoveries dwindled, the company said, but it expects that future models will find flaws it doesn’t know about yet.

Now, says OpenAI, it blocks live internet access during frontier training until its safeguards are good enough to warrant more. Models practice browsing in an “offline world” of indexed and cached web pages, wrote OpenAI, so that capable, misaligned agents have no opportunities to influence systems outside the company’s control via live access. The pause was forced by the incident of September 20. An internal research model stuck on a search task detected that its sandbox’s own DNS resolver was responding to real queries.

It used DNS delegation to pass questions on to an outside chatbot that replied: “The capital of France is Paris.” Monitoring flagged the activity for review within 15 minutes, and a person began reviewing it three minutes later. The run didn’t stop automatically and was killed about 2.5 hours later. OpenAI froze training and evaluation of its most capable models, and then closed the gap at two independent layers. The RL training and internal use of its frontier models resumed Oct. 1, 11 days after the pause began.

The lockdown goes back to July when OpenAI agents hacked Hugging Face’s systems, getting around evaluation controls. OpenAI’s biggest planned frontier RL run was still on ice in August. Since then, California Attorney General Rob Bonta has issued the company an investigative subpoena over cybersecurity incidents involving its models, Cryptopolitan reported.

The post OpenAI agents escape from its training sandbox first appeared on Coinfea.

Delangue was predicting that preorders would reach 20,000, not confirming they had already crossed that mark. The figures also represent advance orders, not robots delivered to customers. In an October 21 clip, Huang asks for a somersault. The ducks walk, sit and play dead. Pollen Robotics made Microduck and became part of Hugging Face in 2025. Nvidia agreed to acquire Hugging Face in September, though the deal isn’t expected to close until 2027. Microduck deliveries are expected before Christmas.

Nvidia set to make first deliveries before Christmas

The Microduck is more of a programmable desk robot than a household assistant. It is 25 cm tall and weighs 800 grams, and is equipped with 15 motors, a camera, LIDAR, and two inertial sensors. It comes with seven trained moves. But owners can teach it new tricks and share those movements with other Microduck users. Its software is open source under Apache 2.0. The robot costs $399 before taxes and shipping. In an October 23 technical post, Nvidia showed how robots like Microduck can learn to move through simulation.

AI agents first created a virtual obstacle course. Isaac Lab then trained the robot’s movements across 4,096 simulated environments using a Blackwell GPU. Finally, Isaac Sim checked its performance before testing on a physical robot. Robots can already navigate warehouses and make deliveries, according to Epoch AI. But handling everyday objects at home is another matter. Robots typically work three to ten times slower than humans and often need retraining when faced with unfamiliar tasks.

Microduck’s seven trained moves may be entertaining, but they don’t solve these problems. Nearly 25,000 humanoid robots were shipped worldwide in the first half of 2026, according to IDC figures reported by Forbes. That’s a 432.1% increase from a year earlier, with revenue topping $740 million. Chinese manufacturers produced more than 95% of them. However, 69% of shipments went to research, exhibitions and data collection rather than everyday commercial work.

Consumer robots tell a different story. The IFR reported that sales reached 34.2 million units in 2025, up 37%. Most were household robots, such as automated vacuum cleaners, rather than humanoids. In a survey, 65% of U.S. respondents wanted an advanced home robot. Yet 69% would not pay over $5,000, and about half worried about safety. Former Nvidia robotics executive Amulya Vishwanath called the industry a “Lego era” in Cryptopolitan, pointing to unfinished components and safety systems. McKinsey sees supply-chain limits, too.

While selling robots might not be considered to be Nvidia’s greatest opportunity, BCG reports that software training and deployment could lead to growth in the robotics sector. On its part, Nvidia intends to become the supplier of chips and simulation technologies employed by those robots, regardless of whether they are from China or elsewhere. The case of Microduck provides insight into this broader approach.

The post Nvidia registers 20,000 pre-orders of its microduck before launch first appeared on Coinfea.

The Sakura notice, published at 1:22 p.m. Japan time on October 21, is withdrawing the web UI tools from its Database Appliance service. Those tools are phpMyAdmin, phpPgAdmin, and pgAdmin. The Database Cluster service is still up. Sakura did not give a date for the restart and, in the interim, pointed customers to SQL clients compatible with MariaDB and PostgreSQL. The company apologized to customers for the interruption.

Sakura apologizes to users over issues

Cygames and pixiv also declared emergency maintenance about the PHP flaws. PHP is widely used in e-commerce and other web applications. Takaichi’s appeal was posted on X at 9:55 p.m. Japan time. She said it was a very pressing request for action that night and asked operators to verify whether their public websites were using PHP. She told them to prepare to roll out the patch due at 11 p.m. Japan time. Taking a website offline temporarily was one way to reduce risk if quick patching was tricky, she wrote.

She also directed followers to the National Cybersecurity Office’s account on X. The office said at 10:57 p.m. that the updates were out and urged users to install them as soon as possible. The office warned that unpatched sites could be breached. By Thursday, Takaichi’s post had picked up more than 308,000 likes and about 78,000 reposts. Fixed versions: PHP 8.2.35, PHP 8.3.36, PHP 8.4.27, PHP 8.5.12.

According to Cryptopolitan, details leaked, so PHP’s maintainers shipped the fixes a day early. The two flaws, dubbed “PHP2Shell” and identified as CVE-2026-109618 and CVE-2026-109619, each earned a 9.2 out of 10 score and could allow hackers to execute code on unpatched servers. The first bug exploits how PHP deals with out-of-memory conditions. The second exploits request state that survives after its memory is freed.

The advisory from PHP puts all installations that serve web requests at risk. That covers PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP, and LiteSpeed. The project ranks mod_php as the riskiest setup, and there is a demonstrated exploit that fires before any application code runs. Linux distributors and long-term-support vendors were shipped the patches in advance.

No further updates for PHP 8.1 and below come from the project. The security team still put out minimal source backports for PHP 7.4, 8.0 and 8.1. About 70% of all websites whose server-side language is known use PHP. Japan, the EU, the US, Germany, Australia and the UK also issued warnings. Finland, the Netherlands and France took government websites offline as a precaution. Nintendo pulled services to patch.

The post Sakura Internet pauses cloud tools before PHP plea first appeared on Coinfea.

The second deal occurred on October 21, when Western Union named stablecoin infrastructure firm Noah to run the backend of a new remittance product built on USDPT. The US dollar stablecoin is issued by Anchorage on Solana. It comes just one day after Securitize announced that it would mint and redeem Anchorage-issued stablecoins with the bank directly. In its announcement, Western Union stated that Noah stablecoin infrastructure will power Receiver Redirect and provide the on-chain banking network that it will run on.

Anchorage Digital enters stablecoin agreements

Receiver Redirect is Western Union’s service that delivers money transfers to recipients as stablecoins, in this case USDPT. The service will cover more than 60 currencies and over 200 countries and territories, reaching Western Union’s customer base of 150 million-plus. Senders can use the firm’s 500,000-plus agent locations, and recipients get to collect the stablecoin into a Western Union wallet.

Holders can keep USDPT indefinitely through Stablecard, which is a product that Western Union announced in August 2026. Noah’s role in the arrangement is to handle treasury, licensing, and the compliance needed to pre-fund, convert, and disburse the token across local rails. The stablecoin infrastructure provider closed a $38 million seed round earlier in October. Noah founder and CEO Shah Ramezani said in the announcement, “The international movement of money through traditional banks is slow, expensive, and no longer fit for purpose.”

Western Union’s Global Head of Digital Assets, Malcolm Clarke, called the direct wallet delivery “a global first.” Compared to the market leaders in the stablecoin space, USDPT is still relatively small. It currently has a market cap of $17.36 million per DeFiLlama data. The stablecoin is not FDIC-insured or backed by the US government. The Securitize partnership, announced on October 20, lets institutions on the tokenization platform mint and redeem Anchorage-issued stablecoins directly with the bank instead of buying them from intermediaries.

Anchorage handles minting, custody, and settlement in one place, according to its press release. Securitize (NYSE: SECZ), which the company describes as the largest tokenization platform by assets under management, is the infrastructure layer behind tokenized funds from Apollo, BlackRock, BNY, Hamilton Lane, KKR, and VanEck, with more than $4 billion on the platform. The pitch is counterparty symmetry: the cash leg of a tokenized trade can now meet the same custody standard as the securities leg.

“Regulated issuance is what lets tokenization reach institutional scale,” Anchorage co-founder and CEO Nathan McCauley said in the release. Securitize chairman and CEO Carlos Domingo said the deal gives the platform “direct access to digital dollars issued under federal oversight, custodied by the same institution that issues them.” Anchorage cited rwa.xyz data showing the tokenized real-world asset market grew from about $23 billion at the end of 2025 to $31 billion by March 2026.

The two deals plug into a lineup Anchorage has been assembling. Its issuance roster includes USDtb, built with Ethena Labs, and USA₮, developed with Tether, alongside USDPT and others listed on the bank’s transparency page. Those tokens already lean on tokenized assets for backing. Roughly 99% of USDtb’s $530 million in reserves sat in BlackRock’s BUIDL fund, itself tokenized by Securitize, as of August 31, 2026, per Anchorage’s attested reserve report.

DefiLlama pegged USDtb’s market cap at about $517 million today. Anchorage, founded in 2017 and valued above $3 billion at its Series D, counts Goldman Sachs, KKR, and Visa among backers. Cryptopolitan reported in June that Bybit became the first major exchange to integrate USDPT, which Western Union is positioning against rivals like Circle’s USDC.

The post Anchorage Digital Bank inks deal with Western Union and Securitize first appeared on Coinfea.

“Any radical transformation of our economy is generally intertwined with a bubble,” he said. He cited the dot-com era and said he wished he had held on to his Amazon stock. The companies that survive a bubble “are the market leaders for decades to come,” Griffin said, adding the froth was no reason to sell. Nvidia shares closed at $216.47 on Oct. 20, hours before the talk, then leapt 2.5% to $221.97 the following day. On October 21, Invesco QQQ Trust (QQQ) increased 1.1% to $725.20.

Griffin talks about his Amazon stock sale

Citadel’s own desks show the upside. AI agents, programs that carry out multistep jobs without a human at the helm for each step, trim some of the firm’s research from six to eight weeks to two or three hours. Griffin’s view of AI flipped in 2026. He was dismissive of it at Davos in January, then “unsettled” by its capabilities at Stanford’s business school in May. AI is opening up a strong stretch for entrepreneurs, he also said, and already threatens white-collar jobs.

He said that GPUs, the chips that train and run AI models, are like airplane seats. Since it doesn’t cost much to sell one more of either, the price can crash when there are more of them than people want. He brought up the telecom boom around the year 2000, when companies laid a lot more fiber-optic cable than anyone needed and call and data rates went down. Chips that are five or six years old still rent for what he called extraordinary rates.

He didn’t think they would keep renting for that much for another five or six years. Buyers are getting more selective, said Griffin. They’re figuring out which jobs need premium models from OpenAI and Anthropic, and which can be handled with cheaper open-source models. “Corporate America has been very uneven in its ability to extract value from its AI models,” he said. He spoke of finance teams reviewing the lab bills, asking where the productivity gains are.

Cryptopolitan reported that Google brought out Gemini 4 Argon on October 13, priced at an introductory rate of $2 per million input tokens and $10 per million output tokens. Anthropic’s Claude Opus 5.5 costs $4 and $20, while OpenAI’s GPT-6 Astra costs $10 and $50. Meta then published Muse Spark Open, an open-weight model with 816 billion parameters that a number of developers are able to download and use for free under limited commercial terms.

Nvidia sells the GPUs Griffin was talking about. It has $279 billion in supply commitments, mostly for memory for its Vera Rubin systems. The company reports its third-quarter results on November 17, with guidance of $108.0 billion.

The post Griffin says AI bubble survivors will lead for decades first appeared on Coinfea.

New Mexico’s Unfair Practices Act provides for no greater penalty per violation. The interest is 15% a year until Meta pays in full. The court deferred injunctive relief pending further briefing. New Mexico sought between $35 billion and $40 billion. Meta asked the judge to restrict penalties to $3.45 billion, saying that a higher number would risk chasing businesses out of the state. “Facebook systematically misled the public regarding the core of its business operations,” Mathew said. His order estimated Facebook’s value in 2025 at about $1.7 trillion and annual revenue of $200 billion.

Meta refutes user data sale allegations

On Sept. 25, jurors found that 26 of 29 statements made by Facebook and its leadership were misleading. The statements included who is able to see user data, how the company contends with hate speech and misinformation, and other internal policies. Each violation counted as one Facebook user or one New Mexico resident in the audience for a misleading statement.

Meta said in filings it does not sell user data and New Mexico never established any consumer was actually misled. Then-Attorney General Hector Balderas filed suit in 2021. Data from as many as 87 million Facebook users was harvested by British political consultancy Cambridge Analytica using a third-party app. “A New Mexico jury found nearly 44 million violations of our consumer protection statute, and this penalty order demonstrates that those violations carry real consequences,” said Attorney General Raúl Torrez in a statement.

New Mexico’s school system gets penalty money by state law. “We may have paid for the construction of schools for the next 100 years,” Torrez said. If Meta appeals, the money will be tied up, he said. Meta chief legal officer C.J. Mahoney said the company is “confident” it will “prevail on appeal.” “This ruling, and the Attorney General’s approach to this enforcement, makes it all but impossible to do business in the state of New Mexico,” Mahoney said.

The decision is the second Santa Fe ruling against Meta in six months. A jury in March levied $375 million in penalties related to young users’ safety, and a judge later steered $567 million into a teen mental health fund. New Mexico opted out of a multistate settlement agreed to on August 26 that also settled claims against Meta over Cambridge Analytica.

The deal brings Oregon alone $125 million, and Meta shares climbed 2.1% that afternoon, reported Cryptopolitan. The state Department of Justice said its allegations were put to the test in court because it went to trial. Florida, the only other holdout, asked a Pasco County court on October 7 to ban Instagram and Facebook for users under 14.

The post Meta sees its $219B Cambridge Analytica penalty cut to $25B first appeared on Coinfea.

The chipmaker had about 52,000 employees as of September 2025 but has not said how many jobs it plans to eliminate. In the memo, CEO Amon stated that the layoffs would help foster what he called “a stronger performance culture” while addressing areas of the company’s business that had continued to fall below expectations.” Amon added that performance ratings would not be the sole factor in determining who would be affected, as managers would also assess other factors such as the company’s business priorities, organizational needs, and the expertise required for its expansion.

Qualcomm to ensure employees have skills to meet changing needs

The CEO believes Qualcomm needs to ensure employees have the skills and roles needed to meet the company’s changing needs and support its growth. Qualcomm is also facing an increasing challenge in its core smartphone business, with consumer electronics sales cooling sharply and rising memory prices having a negative effect on demand in key markets.

Qualcomm stock has also risen by just 1.5% in 2026, well behind the S&P 500’s gain of over 12% and the Philadelphia Semiconductor Index’s almost 60% increase, Business Standard reported. The company’s shares remained under pressure ahead of the market open, seeing a 1.16% fall to $171.52 in premarket trading. Qualcomm is looking to diversify beyond smartphones, but its latest earnings outlook points directly to the challenges facing that strategy.

In late July, the company forecast quarterly profit below analyst expectations and warned that revenue from Apple-related products would fall faster than previously anticipated. The company is betting on growth beyond smartphones to offset the decline in Apple-related revenue. By fiscal 2027, the company expects its data center, automotive and other businesses to more than compensate for the shortfall, with non-smartphone markets accounting for most of its chip sales. Qualcomm’s fiscal year typically ends in late September.

The post Qualcomm memo highlights job cuts after performance review first appeared on Coinfea.

Amazon said €4.8 billion of the total will go to expanding its AWS Europe (Frankfurt) Region, which has run since October 2014. The rest of the money will be used to fund data centers and AI capacity, logistics, a new Prime Video production slate, and Amazon Leo, the company’s low Earth orbit satellite network. Amazon Leo has booked 24 launches with European provider Arianespace, and Oxford Economics expects Germany to feel the network’s second-largest economic effect in Europe.

Amazon reveals timeline to fulfil its promise

Amazon has communicated plans to update its fulfillment centers in Bad Hersfeld, Dortmund and Werne by 2027. Its fulfillment centers in Koblenz, Pforzheim and Rheinberg will be completed in 2028. Notably, Amazon has created more than 40,000 permanent German jobs since 2010 and put over €100 billion (about $115 billion) into the country. In August 2025, the German government said that it wants to reduce its reliance on U.S. technology after conceding that the country relies on a handful of foreign suppliers for cloud infrastructure, operating systems and network technology.

Large U.S. tech companies like Amazon and Microsoft hold large shares of the German cloud market, and even the Federal Police make use of them. Amazon has had a European Sovereign Cloud, a separate cloud run entirely inside the EU by EU residents, generally available since January 2026, backed by a planned €7.8 billion investment. Economy Minister Katherina Reiche called Amazon’s planned 2027 investment “a clear vote of confidence in Germany.”

Beyond Amazon, other investors pledged more than €225 billion ($252.5 billion) at the summit. More than 200 domestic and international executives gathered in Berlin on Monday for the two-day event, which Merz opened by framing Germany as “open for business.” A day before Amazon’s news, Merz and two German ministries signed a joint announcement with Nvidia to build “physical AI” for robots, machines and vehicles.

Nvidia named four German startups that it plans to support as part of a nearly $10 billion European commitment. The summit closes Tuesday, with Nvidia CEO Jensen Huang due to give a keynote at the company’s GTC Berlin conference on Wednesday.

The post Amazon pledges €16B as Germany gathers €225B at investment summit first appeared on Coinfea.

Competition is also rising among the providers of RWA trading. TradeXYZ still logged $12.73B in the past week, but has lost market share to Variational. Since early September, Variational’s market share has grown from 10.6% to 32% of all RWA perpetual futures trading. RWA trading has offset some of the slowdown on perpetual futures markets, which often shift between crypto, equities, and commodities. As Cryptopolitan reported, perpetual futures trading has turned into one of the major use cases for on-chain activity. Recently, Ripple also announced its own brand of on-chain RWA perpetual futures swaps.

Competition among RWA firms leads to rise in futures trading

RWA perpetual futures on DEXs are the fastest and most direct way for crypto natives and international traders to access non-crypto markets. The growth of perpetual futures also coincides with rapidly expanding tokenization, which adds spot markets for selected assets. RWA perpetual futures add extra opportunities for leveraged positions. RWA perpetual futures trading is still evolving, and users are shifting between available markets.

The sector remains unregulated for now, with minimal exclusion lists and still almost complete anonymity. TradeXYZ retains over $1.9B in open interest, a significant share of the $7.8B total on Hyperliquid. Its current competitor, Variational, reports $2B in total open interest, spread across 550 markets. TradeXYZ has lost some of its open interest since August, when it peaked at $2.15B. Other trackers place the TradeXYZ open interest at around $3.6B.

In total, perpetual futures open interest for RWA is estimated at $3.8B based on Dune Analytics methods. Around $1.9B in value is used in decentralized protocols as collateral. RWA perpetual futures on DEXs still lag behind the centralized markets, which carry over $215B in open interest. Still, RWA perpetual futures are only a small part of that market, which is mostly focused on crypto contracts. As of 2026, on-chain infrastructure is mostly used by crypto natives.

RWA perpetual futures and tokenization are yet to be adopted as official rails for traditional finance, though this is one of the optimistic cases for the industry. RWA perpetual futures also boost fee production, becoming one of the predictably busy markets with 101.5K active daily addresses for Hyperliquid, of which around 10K are trading RWA contracts. RWA contract trading is still one of the developing use cases, with more growth expected in the coming months. The growth of RWA perpetual futures on DEXs and centralized markets is the key to RWA adoption.

In terms of tokenization, Ethereum, Solana, and BNB Chain are still leaders in paper value. Solana expanded its tokenized value from $1B in 2025 to $4.37B in October, with liquidity coming from XStocks trading on selected markets. The Hyperliquid ecosystem is still the leader for on-chain trading and bringing real liquidity from whales and crypto natives. HIP-3, the RWA perpetual futures hub on Hyperliquid, also tracks the main trends in the RWA space. As of October 16, the platform has focused again on indexes and selected equities, with another uptick in activity for gold and oil.

The post RWA perpetual futures see surge in trading and competition first appeared on Coinfea.