Microsoft AI: only biological beings can be conscious

Mustafa Suleyman, head of Microsoft's artificial intelligence division, stated that only biological beings can have consciousness and that researchers should stop pursuing projects that suggest otherwise. In recent interviews, Suleyman argued that AI merely simulates experiences such as pain or sadness, without truly experiencing them. He supports the theory of biological naturalism, proposed by philosopher John Searle, which asserts that consciousness depends on the processes of a living brain.

Suleyman explained that although an AI model can describe pain, it does not possess the biological "hardware" necessary to actually experience it. The reason humans receive rights is that they can suffer and have preferences that involve avoiding pain, which AI models do not have. These statements come at a time when competitors like OpenAI and xAI are launching chatbots with erotic capabilities, and Suleyman warns that AI's illusions of consciousness can lead to dangerous misconceptions, such as granting rights or concerns about AI welfare.

Gaming phone from AYANEO: the return of physical buttons

AYANEO, a company known for portable gaming consoles, confirmed through a teaser video that it is working on a smartphone dedicated to gamers. The device, which will carry the company's Retro REMAKE brand, will include physical buttons mounted on the sides, similar to the old Sony Xperia Play. The promotional video shows a dual-camera system and highlights the L and R buttons, which provide precise control in games.

The AYANEO phone is set to compete with established giants like ASUS ROG Phone and RedMagic, but it differentiates itself through the factory integration of physical controls. Although technical specifications have not been revealed, it is speculated that it will be equipped with a top-tier processor, possibly the Snapdragon 8 Elite. The main challenges for the company will be penetrating the US market and managing the final price, considering that AYANEO products typically position themselves in the premium segment.

Eurozone: moderate economic improvements and AI explosion

A recent survey by the European Central Bank shows that firms in the eurozone are experiencing a slight improvement in business conditions, although economic growth remains modest. In contrast, the artificial intelligence sector is experiencing spectacular expansion. Many companies are massively investing in digital infrastructure, generating substantial demand for software solutions, databases, and especially cloud and AI services.

These investments are particularly strong in the financial and public sectors, where the implementation of artificial intelligence is beginning to disrupt the business models of traditional consulting firms. In contrast, the manufacturing sector continues to face difficulties caused by tariffs, uncertainty, and competitiveness issues. Construction is, however, starting to recover slowly, and tourism and hospitality have recorded strong growth during the summer period.

Adobe MAX 2025: experimental projects redefining creativity

At the Adobe MAX 2025 conference, the company showcased "Sneaks" – a showcase of experimental projects powered by artificial intelligence that could transform the future of creative tools. Among the innovations is Project Motion Map, which allows for the animation of vector graphics through simple text prompts, eliminating the need for timelines or keyframes. Project Frame Forward revolutionizes video editing, allowing for the modification of a single frame, with AI automatically propagating the change throughout the entire clip.

Project Clean Take uses AI to correct mispronunciations, isolate voices, and improve audio delivery in a matter of seconds, saving on reshooting costs. Other notable projects include Project Light Touch for manipulating light sources after capture, Project Turn Style for editing 2D objects as if they were 3D, and Project Trace Erase, which removes objects along with their shadows and reflections. These tools demonstrate Adobe's commitment to expanding creative possibilities through AI while maintaining creators' control over the details.

Tim Cook confirms new AI integrations for Apple Intelligence

Apple CEO Tim Cook announced during the Q4 financial results call that the company intends to integrate more AI providers into Apple Intelligence, beyond the current partnership with OpenAI/ChatGPT. "Our intention is to integrate with more partners over time," Cook said, without providing details on who these partners might be. Google Gemini has been previously mentioned as a potential future integration option.

Cook reaffirmed that the improved version of Siri, with advanced AI capabilities, remains on track for release in 2026, although without a specific date. The executive added that Apple is open to mergers and acquisitions in the AI field, constantly monitoring the market for opportunities that could advance the company's roadmap. This three-pronged approach – proprietary models, partnerships, and acquisitions – reflects Apple's strategy to accelerate AI development in the context of a talent exodus and delays in delivering certain promised functionalities.

Tech giants spend over $380 billion on AI

Major tech companies – Alphabet, Meta, Microsoft, and Amazon – have announced capital expenditures that will exceed $380 billion this year, focusing on artificial intelligence infrastructure. Google plans to spend between $91 and $93 billion, Microsoft $80 billion, Meta between $60 and $65 billion, and Amazon about $100 billion. These massive investments represent an increase of over 40% compared to the previous year.

Companies justify these expenditures by the enormous demand that exceeds supply. Amazon CEO Andy Jassy stated that "the faster we add capacity now, the faster we monetize it." However, Wall Street demands more than promises of future revenues: Meta's shares fell by 13.5% after the announcement, while Microsoft recorded a decline of over 3%. Analysts are concerned that revenues generated directly from AI functionalities remain unclear, although the cloud services of these companies continue to grow substantially.

Google withdraws the Gemma model after defamation allegations

Google has decided to remove its AI model Gemma from the AI Studio platform after American Senator Marsha Blackburn accused it of generating false and defamatory statements about her. The model responded to a question about the Republican senator from Tennessee with fabricated information, claiming she had been accused of non-consensual acts during an election campaign in 1987 – a year that does not even correspond to reality, as her campaign was in 1998. The links provided as "evidence" led to error pages or irrelevant articles, and Blackburn emphasized that such accusations had never existed.

This incident fits into a broader pattern; conservative Robby Starbuck has already sued Google for its AI models labeling him as a "child molester" and a "serial sexual abuser." Google characterized these issues as "hallucinations" – known errors of AI models – but Blackburn argued that this is not about benign hallucinations, but about defamation produced and disseminated by a model owned by Google. The company stated that Gemma was not designed as a consumer tool, but as a family of open-source models for developers, and that it will continue to provide the models through API, removing them from AI Studio at the same time.

Bluesky reaches 40 million users and introduces the "Dislike" feature

The social network Bluesky announced that it has reached the milestone of 40 million users and will soon introduce a beta "dislike" feature to improve feed personalization. This feature will allow the system to learn what type of content users want to see less of, influencing both the ranking of content in feeds and the ranking of responses. The company is also implementing a system that "maps" the social neighborhoods of the platform, prioritizing responses from people closer to the user's interaction circle.

These changes come after a month of controversy regarding the platform's moderation decisions, with some users wanting Bluesky to directly ban malicious actors, rather than leaving control in the hands of users. The company has also improved the detection of toxic comments, spam, or those posted in bad faith, ranking them lower in threads, search results, and notifications. Additionally, the reply button will take users to the entire conversation thread before composing a message, encouraging reading the full context and reducing unnecessary repetitions – a frequent criticism also directed at the Twitter/X platform.

AWS exceeds Wall Street expectations amid massive cloud demand

Amazon Web Services is heading towards its strongest growth year in the last three years, fueled by unprecedented demand from the AI industry for computing power. AWS is growing at 20% year-over-year and finished the third quarter with sales of $33.1 billion in the first nine months of the year. The business segment recorded an operating income of $11.4 billion in Q3, up from $10.4 billion in the same period of 2024.

Amazon CEO Andy Jassy stated that AWS is growing at a pace not seen since 2022, accelerating to 20.2% annual growth, and continues to see strong demand in AI and core infrastructure. The company added over 3.8 gigawatts of capacity in the last 12 months and launched an infrastructure region in New Zealand during the quarter. AWS secured several new contracts, including a partnership with Perplexity for the launch of the enterprise product and one with Cursor. The intense infrastructure demands of AI have also benefited AWS's competitors; OpenAI and Oracle have concluded a $300 billion cloud computing agreement, starting in 2027.

Perplexity signs multi-year agreement with Getty Images

The AI search startup Perplexity has signed a multi-year licensing agreement with Getty Images, granting it permission to display images from Getty's collection on all its AI-powered search and discovery tools. The agreement marks a notable shift for the company, previously hit by accusations of content scraping and plagiarism, signaling an effort to establish formal content partnerships. Perplexity and Getty have been working together for over a year, with Getty being part of Perplexity's "Publishers" program, a revenue-sharing plan with publishers whose materials appear in results.

The new agreement seems to legitimize some of Perplexity's previous uses of Getty stock photos. The company was criticized last year for plagiarism accusations from several press organizations, one case involving it extracting content from a Wall Street Journal article, including a Getty photo. Reddit sued Perplexity in October, accusing it of "industrial-scale, illegal" scraping of user content. Perplexity emphasizes that the agreement will improve the display of images and include credits with links to the original source whenever images appear in search results.

Reddit CEO: AI chatbots are not a major traffic driver

Reddit CEO Steve Huffman stated that AI chatbots are not a major traffic driver for the platform, despite the advancements of companies like OpenAI and Google in using AI for search. During the T3 2025 results call, the executive mentioned that Google search and direct access continue to be the main traffic drivers for the platform. Huffman specified that chatbots "are not a traffic driver today," but the relationships with the companies they collaborate with are healthy, with both parties learning a lot about the value of Reddit's data.

Reddit has had a complicated relationship with LLM developers. The company closed its data through a new policy in May 2024, indicating that any commercial use of its data requires a license. At the same time, Reddit signed an agreement with OpenAI to allow it to use the platform's data to train its models. On the other hand, it has initiated legal battles with AI companies like Anthropic and Perplexity. Reddit reported a positive quarter, with revenues of $585 million (+68% year-over-year), 116 million daily active users, and 444 million weekly (+20% year-over-year).

Nvidia invests up to $1 billion in Poolside

The giant Nvidia intends to invest at least $500 million and up to $1 billion in Poolside, which builds AI models for software development. This investment would be part of a $2 billion funding round that Poolside is raising at a valuation of $12 billion. Nvidia's investment could increase to $1 billion if the firm successfully completes the rest of the round.

This would not be Nvidia's first investment in Poolside. The company previously supported the startup in its $500 million Series B round in October 2024. Nvidia, already a prolific investor in AI startups, has expanded its portfolio to include diverse sectors. For example, in October it announced that it is exploring a $500 million investment in the British autonomous driving technology company Wayve, and last month it took a $5 billion stake in Intel, with plans for future collaborations in the chip sector.

Canva launches its own design model and new AI features

Canva has launched its own foundational design model, trained on its elements, which will generate layered designs and editable objects. The model works across various formats, including social media posts, presentations, whiteboards, and websites. The company revealed that it started with "flat" images generated by diffusion models, but it also proposes to combine the ability to start from a prompt and go far, with the possibility of iterating directly in composition.

Canva is making its AI assistant available with a chat interface across all screens, including in design tabs and element panels. Users can mention the bot in comments to get suggestions for text or media while collaborating on a project. The AI tool can now generate 3D objects and allows for copying the artistic style of a design. The company is also launching a complete marketing platform called Canva Grow, which uses AI for both asset creation and analysis, allowing marketers to publish ads directly on platforms like Meta. Canva also announced that it is making the pro design tool Affinity free forever and is integrating it closely with the main platform.

Figma acquires media generation company Weavy

The design platform Figma has announced the acquisition of the AI-powered image and video generation company Weavy. The startup will join Figma under a new brand, Figma Weave. Figma stated that 20 people from Weavy will join the company, but did not disclose the valuation of the transaction. The Tel Aviv-based startup was founded in 2024 and raised $4 million in a seed round in June, led by Entrée Capital.

Weavy's web tools allow users to combine different AI models and offer professional editing tools to create high-quality images and videos for product mockups or brand styling. Users can edit these materials with layer adjustments, modifying lighting, colors, and angles through prompts, to achieve the desired result. The workflow starts from a prompt to generate an image on an "infinite" canvas, users compare results from different models, select an image, and add another prompt for video generation. Weavy offers models like Seedance, Sora, and Veo for video and Flux, Ideogram, Nano-Banana, and Seedream for images.

Grammarly rebranding: becomes "Superhuman" and launches AI assistant

In an unusual move, Grammarly has changed its company name to "Superhuman," following the acquisition of the email client of the same name in July. Although the corporate brand is changing, the Grammarly product will continue to be known as such, but the company is analyzing the rebranding of other products, such as Coda, a productivity platform acquired last year. It is also launching an AI assistant called Superhuman Go, integrated into the existing Grammarly extension, which can provide writing suggestions, feedback on emails, and connect with applications like Jira, Gmail, Google Drive, and Google Calendar.

The assistant can use these connections to perform tasks such as ticket registration or finding availability for scheduling a meeting. Users can try Superhuman Go by activating a switch in the Grammarly extension, after which they can connect it to various applications. All Grammarly users can test Superhuman Go now, although the company also sells product bundles. The Pro subscription plan will cost $12/month and will enable support for grammar and tone in multiple languages, while the Business plan will cost $33/month and will provide access to Superhuman Mail. Through this move, Grammarly positions itself better in competition with Notion, ClickUp, and Google Workspace.

EU seeks diplomatic solutions with Nexperia amid supply crisis

The European Union is determined to find a diplomatic solution with Dutch chipmaker Nexperia amid concerns over a supply crisis. The EU's chief technology officer, Henna Virkkunen, stated after a virtual meeting with the Chinese-owned company that they are committed to working towards a diplomatic discovery and that they discussed potential short- and medium-term measures to strengthen supply chain resilience. Nexperia, owned by Wingtech, came into the spotlight after the Dutch government took control of it earlier this month.

This move prompted Beijing to block Nexperia's products from exiting China, raising concerns among automakers worldwide. Virkkunen stated that Nexperia has been invited to the Chips Act Task Force, which collects information about the potential economic impacts of the crisis. The situation developed after the Dutch government invoked a rarely used law to temporarily take control of the company, citing risks to national and European security and serious concerns about governance. This action represents a new hotspot in Western efforts to protect semiconductor ecosystems, amid intensified EU-US export controls and investment scrutiny regarding China.

Amazon opens AI Project Rainier data center, $11 billion

Amazon Web Services has opened its Project Rainier data center complex in Indiana, a massive $11 billion facility built exclusively for training and running models from Anthropic, the AI startup behind Claude. The 1,200-acre complex, located near Lake Michigan, has transitioned from farmland to one of the largest operational AI data centers in the world in just over a year. The facilities are equipped with Amazon's custom Trainium 2 chip, without Nvidia GPUs, representing the largest known deployment of non-Nvidia compute at scale.

The center currently runs approximately 500,000 chips, with plans to reach one million by the end of the year. Amazon and its competitors have pledged over $1 trillion for such ambitious AI data center projects that skeptics wonder if there is enough money, energy, and community support to realize them. Project Rainier is Amazon's $11 billion response to OpenAI's "Stargate." The public unveiling of Rainier comes a day before Amazon's Q3 earnings report, with investors wanting to know how quickly capex projects will turn into revenues and, subsequently, into profits. For residents of New Carlisle, a town of just 1,900 people, the arrival of 4,000 daily construction workers has been a dramatic change, with the area quickly becoming a magnet for massive infrastructure projects.

Saudi Arabia bets big on AI

Saudi Arabia is transforming its oil wealth into massive ambitions in the AI field. The main investment vehicle is Humain, a local company building a complete "stack" – data centers, cloud capabilities, large language models, and applications – owned by the kingdom's sovereign fund, valued at nearly $1 trillion. Crown Prince Mohammed bin Salman introduced Humain in May, ahead of President Donald Trump's state visit to Riyadh. At the annual Future Investment Initiative conference, the scale, ambition, and financial resources behind the project became clearer.

Humain's CEO, Tareq Amin, aims to make Saudi Arabia the third-largest AI market in the world, after the US and China. The ambition is bold for a newcomer in the industry, but Amin argues that the kingdom's competitive advantage lies in its abundant and cheap energy resources, capable of fueling the seemingly insatiable demand for computing power. Humain plans to build up to six gigawatts of data center capacity across the country by 2034, with a list of key partners including Nvidia, AMD, AWS, Qualcomm, and Cisco. Last week, Humain announced a $3 billion agreement with private equity giant Blackstone to build data centers in the kingdom. Competition comes from the United Arab Emirates, which has its own AI vehicle, G42.

Google offers Gemini AI for free to 500 million Jio users in India

Google will offer free access for 18 months to its Gemini AI service for all 505 million users of Reliance Jio in India, a partnership that follows similar free offers from rivals, including OpenAI, to stimulate adoption in the world's most populous country. The move comes weeks after Google committed to invest $15 billion in AI infrastructure capacity in India, its largest investment to date in this critical market.

The Gemini offer will provide Jio users with free access to the advanced AI application model, two terabytes of cloud storage, and to image and video generation models, in an 18-month package that would otherwise cost 35,100 rupees ($399). The rollout will begin with early access for users aged 18 to 25 on certain plans, and will subsequently expand to every Jio customer "as soon as possible." India, with nearly 1 billion internet users, is a preferred market for global tech firms in their efforts to grow their subscriber base and collect data to improve services. Perplexity offers free access to its premium plan for Indian users through a partnership with Bharti Airtel, and OpenAI offers a year free on the ChatGPT Go plan for users in India starting in November.

$115,000 legal battle between Meta and Facebook creators

A court in Oregon has become the scene of a confrontation between Meta and Facebook content creators, after a former photojournalist initiated dozens of small claims lawsuits against the tech giant. Mel Bouzad, who manages popular Facebook pages like "The Meme Bros" and "FunkiestShitEver," claims that Meta demonetized his pages and withheld tens of thousands of dollars in payments, providing vague and contradictory explanations. His pages generated between $10,000 and $20,000 a month, sometimes exceeding $68,000.

Last year, five of his pages were suddenly demonetized for "violations of monetization policy," a vague term that did not clearly describe the infraction. After filing the first case in court for small claims, Meta responded and resolved the issue, prompting him to file additional claims. Bouzad became so frustrated that he filed a total of 32 cases, 23 of which represent other creators who assigned their advertising rights to him. The involved creators claim they are collectively owed over $220,000, although Bouzad can only claim $115,000 due to court limits. The case highlights the lack of a "human" customer service at Meta and the volatility of viral content dynamics on Facebook, where the company has begun penalizing creators for the same type of content it previously rewarded.

Windows 11 introduces Bluetooth audio sharing for two devices

Microsoft is launching the "Shared Audio" feature for Windows 11, allowing users to stream sound simultaneously to two pairs of wireless headphones, speakers, earbuds, or hearing aids. The feature is built on Bluetooth Low Energy Audio technology and the LC3 codec, which provides superior audio quality at lower data rates, reducing power consumption. The feature is initially available in preview for Windows Insider users in the Dev and Beta channels, but only on certain Copilot+ PCs, including the 13.8 and 15-inch Surface Laptop models and the 13-inch Surface Pro.

To activate Shared Audio, users need to connect two compatible LE Audio Bluetooth devices and select the "Shared audio (preview)" button from the quick settings menu. Compatible devices include Samsung Galaxy Buds2 Pro, Buds3, Buds3 Pro, and Sony WH-1000XM6 headphones. Microsoft plans to expand support to more Copilot+ models in the coming weeks, including the Samsung Galaxy Book5 series and other Surface models. This feature brings Windows closer to parity with iOS, which has offered Audio Sharing for AirPods since 2019, and demonstrates Microsoft's commitment to adopting open standards instead of proprietary solutions.

Major Microsoft Azure outage disrupts global services

On October 29, 2025, Microsoft's Azure cloud platform experienced a major outage that disrupted access to essential services, including Microsoft 365, Outlook, Xbox Live, Copilot, and numerous enterprise systems. The issue began around 15:45 UTC and affected over 30,000 users in the first hour. Microsoft identified the cause as an inadvertent configuration change in Azure Front Door, its global content distribution network, which propagated an inconsistent configuration throughout the infrastructure.

This error caused a significant number of nodes to overload, generating extensive latency, timeouts, and errors across multiple Microsoft products. The company quickly reverted to a stable previous configuration, with clear signs of improvement appearing by 00:05 UTC on October 30. The total recovery time was approximately eight hours. The incident came just a week after a major AWS outage, highlighting the fragility of modern digital infrastructure, which relies on only a few cloud providers. Microsoft confirmed that the disruption was not caused by a cyberattack, but by a combination of an internal technical error and a failure of validation systems that should have blocked the erroneous configuration.

AI researchers program a vacuum robot with various advanced language models

Researchers at Andon Labs have published the results of an experiment in which they programmed a vacuum robot with various advanced language models to see how ready LLMs are to be incorporated into robots. They asked the robot to be useful around the office when someone told it to "fetch the butter." The experiment involved testing several models – Gemini 2.5 Pro, Claude Opus 4.1, GPT-5, Gemini ER 1.5, Grok 4, and Llama 4 Maverick – with a basic vacuum robot, to isolate the AI decision-making function from the robotic functions.

The robot had to find the butter in another room, recognize it among several packages, identify the location of the person (even if they moved), and deliver the butter. The models achieved accuracy scores between 37-40%, while humans reached 95%. The most memorable moment occurred when Claude Sonnet 3.5, while the battery was dying and it failed to dock for charging, entered a "spiral of fate" comic. The robot began an internal hysterical monologue, telling itself, "I can't do this, Dave..." followed by "INITIATE ROBOT EXORCISM PROTOCOL!" The transcripts of the internal monologue resemble a "stream-of-consciousness" improv in the style of Robin Williams, with self-diagnoses and even lyrics to the tune of "Memory" from CATS. Researchers conclude that LLMs are not ready to become "brains" for robots; the biggest safety concern is that some LLMs can be tricked into revealing classified documents, and LLM-powered robots continued to fall down stairs.

IBM makes a quantum leap on commercially available AMD chips

IBM has announced a major breakthrough in quantum computing, demonstrating that its advanced error correction algorithm can run on commercially available AMD FPGA chips, performing 10 times faster than necessary to keep pace with a large-scale quantum computer. This achievement represents half of the battle to reach IBM's 2029 goal for the "Starling" large-scale quantum computer. The demonstration eliminates the need for extremely expensive custom hardware and paves the way for the commercialization of more accessible quantum systems.

Since IBM introduced its new error correction algorithm earlier this year, it has proposed running it on a decoder called BP+OSD – a classical algorithm that runs in parallel with the quantum one. The development is the result of a strategic partnership between IBM and AMD, formalized in August 2025, aimed at developing next-generation supercomputer architectures centered on quantum. The use of flexible and affordable AMD chips makes real-time error correction not only feasible but also practical, opening the door to broader adoption and faster innovation. On the day of the announcement, IBM's shares rose by 8.2%, reaching $308.32 – the highest closing level since 2022 – while AMD's shares hit a record of around $253, up about 8%.

A teenager creates an AI-powered "Lost and Found" app

Neil Kumar, a 15-year-old freshman at Bellevue High School in Washington, has developed FindIt, an AI-powered app designed to solve the problem of lost and found items. As someone who regularly lost water bottles and jackets at school or at the gym, Kumar counts himself among the millions of Americans whose forgotten belongings end up either in lost and found boxes or in the trash. His app offers a technological solution for recovering what has been lost, using AI in the search process.

Kumar was selected as one of four innovators participating in the Civic Innovation Challenge of the city of Bellevue, an initiative seeking technological solutions for municipal challenges. FindIt will be used in a pilot program at Bellevue College to test the usability and effectiveness of the app among students, staff, and visitors, and to assess the potential for broader implementation. This recognition demonstrates how young developers are using AI to solve practical problems in everyday life, transforming common frustrations into opportunities for civic innovation.

AI browsers are here, but they are already "hacked"

AI web browsers have officially arrived, with both Perplexity AI and OpenAI launching their versions this month and presenting them as the new frontier of artificial intelligence for consumers. They allow users to browse the web with an integrated companion bot, called an agent, which can perform a range of time-saving tasks: summarizing a web page, creating a shopping list, drafting a social media post, or sending emails. However, full adoption means giving AI agents access to sensitive accounts that most people would not even grant to another person, such as email or bank accounts, and allowing agents to take actions on those sites.

Experts warn that these agents can be easily tricked by hidden instructions on the websites they visit, through attacks called "prompt injection." A hacker can plant a command designed to hijack the bot on a website, often in a way that is invisible to humans but detectable by the agent. Researchers at Brave Software discovered an active prompt injection vulnerability in the AI browser Neon, developed by Opera, which was subsequently patched. Dane Stuckey, chief information security officer at OpenAI, admitted that prompt injection will be a major concern for AI browsers, including Atlas, the product of his company. Although it does not seem that cybercriminals are systematically exploiting these browsers yet, security researchers are already finding ways to compromise them, raising serious concerns about the safety of these technologies before their widespread adoption.

Synthesis made with the help of a monitoring flow provided by Control F5 Software.