ogle

Google (GOOG.O) announced the official launch of the Claude model deployment option managed by Anthropic on Google Cloud.

OpenAI is bringing its artificial intelligence models and Codex programming tools to Google Cloud, expanding its reach to serve customers of the three major cloud service providers in the United States. Google Cloud customers will purchase OpenAI's cutting-edge models through the "Model Garden and Marketplace," with related features expected to be fully launched in the coming weeks.

Google Cloud and NVIDIA announced that the NVIDIA Vera Rubin infrastructure is officially open for full use.

Kent Walker, a senior executive at Google, dismissed European concerns about the dominance of American companies in the AI industry, stating that the competition between large tech companies and startups is fierce, driving down prices and creating opportunities for European businesses.Walker pointed out that the companies involved in the competition include Nvidia, Meta, SpaceX, and others.

Google's advertising policy announcement indicates that starting from October 21, 2026, advertisers targeting cryptocurrency exchanges and software wallets in the UAE must obtain permission from one of the designated local regulatory authorities and be certified by Google.Acceptable licensing authorities include the Capital Markets Authority (CMA), the Dubai Financial Services Authority (DFSA), the Financial Services Regulatory Authority (FSRA), and the Virtual Assets Regulatory Authority (VARA). Advertisers must also comply with other local legal requirements, and this policy applies to all accounts globally that target the aforementioned advertisements in the UAE.

According to Bloomberg, citing informed sources, former Google CEO Eric Schmidt holds about 1% of shares in Anthropic, stemming from his investment of several tens of millions of dollars in the Series A funding, which will make him one of the individuals benefiting the most from Anthropic's upcoming IPO. It is currently unclear whether he has sold any shares, and both Schmidt and Anthropic declined to comment.According to Pitchbook data, when Schmidt first invested in 2021, Anthropic was valued at $461 million. Investors stated last week that Anthropic's reasonable valuation is between $1.8 trillion and $2 trillion; if calculated at the upper limit, Schmidt's net worth would increase to nearly $80 billion.

Google Research's open-source AI toolchain self-improvement framework Harness² allows AI Agents to automatically modify system prompts, memories, skills, tool rules, as well as scripts and plugins without retraining the model. The project is licensed under the Apache 2.0 license and supports OpenCode and Codex. Its core mechanism enables the Agent to learn through trial and error, proposing several improvement plans when encountering new tasks, which are then executed and compared by different versions of the Agent, ultimately combining to form a more suitable solution.Even without a standard answer, the system can seek improvement directions based on the actual execution process. After the task is completed, the system remembers which modifications were effective and the reasons for them, for reference in similar future problems. Google tested using GLM-5 and Gemini 3.5 Flash. On WorkBuddy-Bench, GLM-5 paired with OpenCode improved its average score from 64.2 to 75.7 after just one round of optimization, surpassing the unoptimized Gemini 3.5 Flash's score of 72.4.In legal task testing, the pass rate for GLM-5's partial scoring criteria increased from 59.6% to 71.4%. In the experiment, Claude Opus 4.8 was defaulted to propose and implement improvements, while GLM-5 and Gemini were primarily responsible for executing tasks, not relying entirely on their own evolution. Domestic developer Joseph Zhou demonstrated an experiment where Harness² integrated with Jev 1.14, claiming that the performance on visual tasks continuously improved.

Google stated that its Willow quantum chip reduces the error rate in preparing the special states (Magic States) required for quantum computing by about 40 times through a method called "Cultivation." The related research was published this week in Science, which is expected to lower the hardware costs of future large-scale quantum computers and draw more attention to the potential threat of cracking Bitcoin wallets.Magic States are a key resource needed to perform certain quantum computing operations, and their preparation process consumes a significant amount of hardware. Google's experiment achieved about 99.99% state fidelity, but only about 8% of attempts passed the screening. Researchers indicated that future experiments need to be scaled up and issues such as real-time operation and error correction need to be addressed. In March of this year, Google estimated that under specific hardware assumptions, future quantum computers might only need fewer than 500,000 physical qubits to crack the cryptography used in Bitcoin and Ethereum wallets.For Bitcoin, some early wallets' public keys have been public since their mining, theoretically making them targets for future quantum attacks; while wallets with undisclosed public keys may face different attack conditions. However, this breakthrough does not mean that quantum computers are capable of cracking Bitcoin. Google's previous cost estimates for attacks assumed the use of efficient Magic States preparation schemes, so this result cannot simply be understood as a 40-fold reduction in the cost of cracking Bitcoin. Research is still in the advancement stage of quantum hardware and error correction technology, and the complete equipment and engineering capabilities required for actual attacks have not yet been realized.

Google announced a breakthrough in the field of quantum computing, stating that its Willow chip has overcome a critical resource bottleneck.Google stated that a technology called magic state cultivation is now running on the Willow chip, reducing errors by approximately 40 times and achieving an accuracy of 99.99%. This research was published in the journal Science, and Google also mentioned that it still needs to prove the effectiveness of this technology under large-scale conditions.

The Google Quantum AI team published a paper in Science, announcing that a technique called magic state cultivation reduces errors by about 40 times on the Willow processor, producing logical fidelity of 99.99% for T states. Magic states are one-time resources required to run useful quantum programs, consumed with each use; according to estimates cited by Google, the cost of producing magic states can account for up to 95% of a quantum computer's total capacity, leaving very limited space for actual computation.The previous standard approach was magic state distillation proposed by physicists Sergei Bravyi and Alexei Kitaev in 2004, which required a large number of qubit factories dedicated solely to this task. Google researchers proposed a cheaper cultivation route in 2024, which behaves more like a filter: injecting rough magic states into a small area of the chip, running a series of checks on them, discarding all attempts found to be defective, and allowing the passers to gradually expand. In the experiments, only about 8% of attempts passed all checks, meaning that more than 90% of attempts had to be discarded for each clean state retained.The cultivation technology is one of the supports for Gidney's 2025 estimate that RSA-2048 encryption could be broken by fewer than a million noisy qubits, a reduction of 20 times from the 20 million estimated in 2019. This experiment used up to 54 of the 105 qubits in Willow, with the paper noting that breaking the RSA-2048 key requires magic states with an error rate of about one in ten billion, which is about a million times cleaner than this result.

Google DeepMind released the first comprehensive empirical assessment of automation in machine learning research and development in the Gemini 4 Argon security report. The conclusion is that AI has not accelerated the R&D of AI itself, despite the fact that AI has written most of the code delivered by the lab's engineers. Researcher Haydn Belfield, who participated in the assessment, stated on X that there is currently no significant end-to-end ML R&D acceleration, so the risks of AI-driven ML R&D are currently low.The team built an internal index, GCI v1, which synthesizes the scores of 123 AI models across 92 benchmarks into a single value. The cutting-edge Gemini capabilities are improving at a stable rate of about 0.1 GCI points per month, with Gemini 1.5 Pro approaching 0.5 points by the end of 2024, Gemini 3.8 Flash slightly above 2.9 points by September 2026, and Argon about 3.3 points above the trend line. However, Belfield noted that this is due to increased computing power and organizational investment, rather than AI-driven acceleration.Over 75% of the code produced by DeepMind is written by AI, and the proportion of fully delegated code merge requests increased from 7.5% to 11.9% between April and September, but about 88% of requests still require multiple rounds of human guidance or manual writing. A survey of 12 senior researchers showed that automation is highest in clearly defined, verifiable execution tasks and lowest in areas such as open science concepts, new experimental designs, and unstable diagnostic training.

According to Variety, CNN has reached a multi-year partnership agreement with Google (GOOG.O) aimed at promoting the application of generative artificial intelligence.

Google has started to roll out Gemini 4 Argon to Google AI subscription users, applicable on platforms including Google Antigravity, the Gemini app, and the AI mode in Google Search. Gemini 4 Argon is launched at a promotional price of $2 per million input tokens and $10 per million output tokens.

Google collaborates with Planet to launch the "Project Suncatcher" prototype satellite through SpaceX Transporter-18.

Security researcher @cyberscrilla discovered that a fake Ledger website and application appeared at the top of Google search results, aiming to trick users into revealing their 24-word recovery phrase. Reportedly, Google data shows that the website had over 1 million visits in the past 30 days.Previously, cybersecurity company Zscaler found in a September investigation that malicious Google ads were impersonating Ledger and redirecting users to a fake wallet verification page. Currently, Ledger is investigating a suspected wallet theft case involving Southeast Asian dealer CryptoBilis, with an estimated amount of $86 million involved.

Google (GOOG.O) announced that Google Cloud has launched the Gemini agent.

According to informed sources, the crypto media company Cointelegraph is seeking to sell, but the amount it is looking for has not been disclosed. The source requested anonymity due to the private nature of the matter. The long-term sluggish and sideways movement of crypto prices has shifted user attention away from crypto news, adversely affecting several digital asset news organizations.Cointelegraph's traffic has been severely impacted by Google. After Google issued a manual penalty in October 2025, the crypto news website's organic traffic dropped by about 80%, and the site disappeared from Google search results. Previously, in June 2025, its website also suffered a front-end vulnerability attack. Data from Similarweb shows that the site had over 12 million monthly visits in December 2024, while as of September 1, the monthly visits were just slightly above 700,000.Cointelegraph was founded in 2013, and its LinkedIn page shows that it has over 200 employees. The company's MENA business was acquired by Luna Media Corporation in July 2022 to support its global and regional expansion. Cointelegraph did not immediately respond to requests for comment.

On October 7, Google launched the web platform Playground, where users can describe game ideas in natural language and obtain playable, shareable game versions without writing code. Users can also continue to adjust physical effects, rules, characters, or scenes through dialogue and share their creations via links. This tool is currently open to users over 18 in the United States, with creation permissions tiered according to the user's Google AI subscription level.Games in Playground can run in browsers on mobile phones and laptops. Creators can set their works to private or publish them to the community gallery, and all games must pass a security review, with some categories supporting leaderboards and multiplayer modes. Google refers to it as "an early experiment," and the announcement did not disclose pricing or specify which AI models power the tool.The upgrade path points to Unity Spark, and Google stated that Playground will soon integrate this tool to meet the needs of creators requiring professional-level mechanics and high-fidelity 3D. In a joint statement, Google and Unity announced that Spark will launch later this year, with closed testing starting soon. Previously, DeepMind released the Genie model in 2024, and in January, Google opened Project Genie based on Genie 3 to users of the AI Ultra subscription in the United States.