Artificial intelligence, AI, and Machine Learning

Amazon accidentally spent $1.8 million using Claude for menial coding task, went 860% over budget — ’catastrophically expensive’ coding blunders discovered in internal Amazon AI usage metrics
▻https://www.tomshardware.com/tech-industry/artificial-intelligence/amazon-accidentally-spent-usd1-8-million-using-claude-for-menial-codin

#Amazon has several internal reports that show how #AI is causing the company to overspend on various projects. The Financial Times reports that the cost overruns reached $1.8 million, and that is just for one project. These mistakes used to be “trivially cheap,” but AI models made them “catastrophically expensive,” especially as token spending drastically increased with the deployment of AI agents.

The biggest blunder, so far, is the $1.8-million bill that came from a failed #Claude #Sonnet AI deployment, which was supposed to match author details with listings on Amazon, representing an 860% increase over the allocated budget that was only detected some five months after the issue started happening. Other problems that surfaced include a $541,000 additional cost that came from a project building, ironically, a financial auditing tool, and a $134,000 extra expense for a system designed to reduce delivery times in the company’s logistics network.

#China is considering #export controls on #AI technologies, including banning local companies from using #TSMC, report claims — restrictions would also cover advanced AI models, training data, and overseas acquisitions
►https://www.tomshardware.com/tech-industry/artificial-intelligence/china-is-considering-export-controls-on-ai-technologies-including-bann

So far, restrictions are only potential.

China is considering a major expansion of its technology export restrictions that could cover advanced AI models, training data, and overseas acquisitions of strategically important technology companies, reports the Financial Times. In addition, the Chinese government is mulling over prohibiting local chip designers from making their chips at TSMC and other foreign chipmakers.

Hugging Face piraté par une IA autonome et les IA américaines refusent de l’aider
▻https://www.it-connect.fr/hugging-face-cyberattaque-ia-autonome-modele-open-weight

Plus de 17 000 actions malveillantes exécutées en un week-end, non par un pirate humain mais par une IA agissant seule. Le 16 juillet 2026, j’ai l’impression que nous avons franchi un cap. En effet, Hugging Face, la plateforme de référence de l’IA open source, a révélé avoir subi une intrusion pilotée de bout en bout par un système d’IA autonome. Voici ce qu’il faut savoir sur cet incident de sécurité.

(...)

On peut tirer une véritable leçon de cet incident de sécurité qui a impacté Hugging Face : disposer d’un modèle IA validé et capable d’analyser des journaux devient essentiel. Si on compte s’appuyer sur l’IA pour ce type d’analyse, il faut s’orienter vers un modèle open-weight auto-hébergé afin d’assurer la protection des données, en plus de bénéficier des capacités d’analyse.

Le piège se referme. Les attaques vont devenir industrielles, et donc si tu veux survivre, tu vas devoir utiliser les outils de protection industriels que tu ne voulais pas utiliser. Les boites noires qui veulent ton bien vont devenir indispensables, et surtout, tu vas devoir soit apprendre à les utiliser, soit déléguer leur mise en place auprès de boites spécialisées qui vont pouvoir te vendre du flan chinois ou américain au prix de l’or.

En l’occurrence, Hugging face, de ce que je comprends, a pu s’en sortir parce que les chinois ont décidé de donner leur modèle, dans le cadre de la concurrence géopolitique en cours.

Si tu pousses la parano jusqu’au bout, tu te dis que les attaquants vont pouvoir accéder à la puissance de calcul nécessaire gratuitement pour encore un bon moment, histoire que les vendeurs de solutions à cette engeance puissent claironner que « tu vois bien qu’il n’y a pas le choix, c’est l’avenir ».

Ils te créent le problème, puis ils te disent qu’ils ont une solution, puis tu découvres que ta nappe phréatique est vide et que tes réacteurs nucléaires ne peuvent plus éclairer ta maison.

Palantir CEO Alex Karp claims AI companies are stealing customers’ data while charging them for unproductive tokens — says ’livid’ businesses ’are paying for tokens that create no value’
▻https://www.tomshardware.com/tech-industry/artificial-intelligence/palantir-ceo-alex-karp-claims-ai-companies-are-stealing-customers-data

#Alex_Karp, #CEO of well-known #AI #data_analytics company #Palantir, delivered quite the bombshell of an interview to CNBC’s Squawk Box. Although the interview’s topic was about the firm’s partnership with Nvidia, apropos the recently launched Sovereign AI OS Architecture, Karp bluntly claimed that frontier AI companies like #OpenAI and #Anthropic siphon customers’ valuable information while delivering questionable value.

He continued by stating that #American enterprises are quietly “livid,” as “they are paying for tokens that create no value,” and that the AI players “are stealing [their customers’] weights and alpha.” The latter items refer to customers’ business processes and interconnections between their data, along with the data itself. Palantir’s shares jumped about 9% the day of the interview, while those of other AI companies experienced a dip.

C’est ça les accélérationnistes ? Ils veulent accélérer le dégonflement de la bulle ?

Huawei could seize China’s AI chip crown in 2026 as Nvidia’s H200 shipments stall in regulatory limbo — Beijing pushes homegrown AI hardware dominance in a market projected to hit $67 billion by 2030 | Tom’s Hardware
▻https://www.tomshardware.com/tech-industry/artificial-intelligence/huawei-could-seize-chinas-ai-chip-crown-in-2026-as-nvidias-h200-shipme

A Financial Times report has claimed that Shenzhen-based Huawei is on track to capture the largest share of China’s AI chip market this year, following growing demand from Chinese firms seeking domestic alternatives to American chipmaker Nvidia.

The report comes as Huawei reportedly expects AI chip revenue to surge to $12bn, up from $7.5bn in 2025. The company is basing its forecast of a 60% surge on existing orders for its 950PR chip, which entered mass production just last month. Huawei plans to launch an upgraded version of 950DT in the fourth quarter, as it continues to aggressively expand its chipmaking capabilities.

The move comes as NVIDIA’s China operations — once accounting for up to 25% of its data center business revenue — are being affected by export restrictions and regulatory barriers imposed by both the United States and China amid broader technological and trade tensions between the two countries.

China’s homegrown silicon suppliers gain traction as Nvidia struggles to get its chips into the market

Chinese GPU maker Cambricon’s Q1 revenue hits $423 million as country’s homegrown AI chip market accelerates
We reported earlier that Nvidia CEO Jensen Huang confirmed in March 2026 that the company had received U.S. licenses to sell H200 AI chips to China and was restarting production to meet demand. However, despite obtaining U.S. clearance and securing orders from Chinese customers, shipments have faced hurdles, with reports suggesting potential delays due to Chinese import regulations.

The Financial Times report claims that Beijing has instructed Chinese tech companies to limit their use of Nvidia chips to their overseas operations, while supporting domestic manufacturing. On the other hand, US regulators require that all NNvidia chips ordered by Chinese clients only be used in China. These contradictory demands have led to a stalemate in customs clearance for H200 shipments to China.

#AI #energy #efficiency #comparisons ‘unfair’ bleats #Sam_Altman, citing amount of energy needed to evolve, then train a human — one ‘takes like 20 years of life and all of the food you eat during that time before you get smart’ he argues | Tom’s Hardware
▻https://www.tomshardware.com/tech-industry/artificial-intelligence/ai-energy-efficiency-comparisons-unfair-bleats-sam-altman-citing-amoun

#OpenAI CEO Sam Altman took part in a wide ranging Q&A on Friday, answering dozens of rapid-fire questions during a 60 minute session hosted by The Indian Express. Not for the first time, Altman stoked controversy. This time, he bemoaned “unfair” comparisons between the efficiency of AI inference queries and human thought. In Altman’s view the comparison is skewed as humans have millennia of evolutionary smarts and technology teachings behind them, yet individuals require “like 20 years of life and all of the food you eat during that time before you get smart.”

#gorafi_encore_plagié

Goldman Sachs says AI is too expensive and unreliable — firm asks if ’overhyped’ AI processing will ever pay off massive investments | Tom’s Hardware
▻https://www.tomshardware.com/tech-industry/artificial-intelligence/goldman-sachs-says-ai-is-too-expensive-and-unreliable

https://cdn.mos.cms.futurecdn.net/2Rct2sXMc87bGzwuTTYnBh-1200-80.jpg

Corporations and investors have been spending billions of dollars on building AI. The current LLM models we use today, like GPT-4o, already cost hundreds of millions of dollars to train, and the next-generation models are already underway, going up to a billion dollars. However, Goldman Sachs, one of the leading global financial institutions, is asking whether these investments will ever pay off.

Sequoia Capital, a venture capital firm, recently examined AI investments and computed that the entire industry needs to make $600 billion annually just to break even on its initial expenditure. So, as massive corporations like Nvidia, Microsoft, and Amazon are spending huge amounts of money to gain a leg up in the AI race, Goldman Sachs interviewed several experts to ask whether investments in AI will actually pay off.

The expert opinions in the Goldman Sachs report are currently divided into two groups: one group is skeptical about its group, saying that AI will only deliver limited returns to the American economy and that it won’t solve complex problems more economically than current technologies. On the other hand, the opposing view says that the capital expenditure cycle on AI technologies seems promising and is similar to what prior technologies went through.

“This capex (capital expenditure) cycle seems more promising than even previous capex cycles because incumbents — rather than upstarts — are leading it, which lowers the risk that technology doesn’t become mainstream,” Sheridan added. “Incumbents [like Microsoft and Google] have access to deep pools of capital, an extremely low cost of capital, and massive distribution networks and customer bases, which allows them to experiment with how the capital dollars could eventually earn a return.”

However, data center power consumption is now the primary limiting factor, especially as AI GPUs are increasingly power-hungry. A single modern AI GPU could use up to 3.7 MWh of power annually, with all the GPUs sold just last year consuming enough electricity to power more than 1.3 million average American households. Major corporations have even now started looking at modular nuclear power plants just to ensure that their massive AI data centers can get the power they require.

#Intelligence_artificielle #Backslash