Les actus

According to Foresight News, SoftBank CEO Masayoshi Son has announced that the company has signed an agreement with OpenAI to increase its investment in the artificial intelligence firm. The total investment, including planned contributions, amounts to $32 billion. SoftBank aims to become the leading global AI agency by the end of this year, with plans for an eventual public listing.

According to Cointelegraph, artificial intelligence activity on blockchains has experienced a substantial increase since the beginning of 2025, driven by heightened funding and user engagement with this emerging technology. Blockchain analytics platform DappRadar reports that AI-related onchain activity has surged by 86% this year, with approximately 4.5 million daily unique active wallets participating in AI decentralized applications (DApps). The rise in daily users has expanded AI app market share from 9% at the start of the year to 19%, closely trailing blockchain gaming, which holds a 20% share. DappRadar analyst Sara Gherghelas emphasizes that the growth of AI is not merely a result of hype but signifies a fundamental shift in user interaction with decentralized applications. AI agents are increasingly serving as a new onchain interface layer, whether through DeFi copilots, social agents, or autonomous gaming assistants. In May, DappRadar predicted that AI agent usage, which involves programs capable of autonomously executing blockchain actions like trading, would soon surpass gaming, a sector that has traditionally dominated the DApp ecosystem. The report highlights that AI agent funding has risen by over 9% in 2025, with $1.39 billion raised by AI agent projects, marking a 9.4% increase compared to 2024. Although this figure is still lower than the funding received by companies like OpenAI, it is noteworthy that AI agent funding now rivals or exceeds other Web3 verticals such as blockchain gaming. Gherghelas suggests that investors in the Web3 space increasingly view AI agents as a new primitive capable of reshaping user interactions with protocols, navigating DApps, and even automating personal financial strategies. She anticipates that 2025 could be the first year AI agents attract more capital than any other Web3 vertical. DappRadar's data from January to June reveals that most AI DApp users are located in Europe, accounting for 26% of all interactions. The largest share of users, at 33%, comes from unspecified regions and users utilizing VPNs or other anonymized sources. Asia closely follows Europe with just under 22% of users, while North America accounts for 15.8%. Gherghelas notes that the global distribution of AI users indicates that AI agents are not a localized phenomenon, with diverse demand spanning continents. Whether managing trades in Asia, representing users in Europe, or interacting with players in North America, AI agents are increasingly becoming a cross-continental presence.

According to Odaily, publicly traded company Done.AI has commenced an evaluation of blockchain infrastructure as a strategic addition to its AI-native financial platform. As part of this initiative, the company plans to allocate 20 million Norwegian Krone (approximately 2 million USD) for the purchase of Bitcoin, with the transaction expected to occur soon. Additionally, Done.AI will explore the feasibility and integration of stablecoins, equity tokenization, and digital securities infrastructure.

According to PANews, British publicly listed company Amazing AI plc, a global fintech group specializing in online consumer loans, announced on June 24 its adoption of a Bitcoin financial policy. This marks a strategic shift towards incorporating Bitcoin into its balance sheet. The company plans to gradually build a Bitcoin treasury using operational funds and upcoming financing. The initial acquisition of Bitcoin is expected to commence in July 2025.

According to Cointelegraph, Elon Musk has announced that his artificial intelligence company, xAI, will undertake a significant retraining of its AI model, Grok, using a new knowledge base devoid of "garbage" and "uncorrected data." Musk revealed in a post on X that the forthcoming Grok 3.5 model will possess "advanced reasoning" capabilities and aims to "rewrite the entire corpus of human knowledge," by adding missing information and eliminating errors. He emphasized the necessity of this approach, citing the prevalence of "far too much garbage" in existing foundation models trained on uncorrected data.Musk has consistently criticized rival AI models, such as OpenAI's ChatGPT, for being biased and omitting politically incorrect information. His vision for Grok is to create an "anti-woke" model, free from what he perceives as damaging political correctness. This aligns with his previous actions, such as relaxing content moderation on Twitter after acquiring the platform in 2022, which led to an influx of unchecked conspiracy theories, extremist content, and fake news. To combat misinformation, Musk introduced the "Community Notes" feature, enabling X users to provide context or debunk posts prominently displayed under offending content.Musk's announcement has sparked criticism from various quarters. Gary Marcus, an AI startup founder and professor emeritus at New York University, expressed concern over Musk's plan, likening it to a dystopian scenario reminiscent of George Orwell's "1984." Marcus criticized the idea of rewriting history to align with personal beliefs, suggesting it represents a dangerous precedent. Bernardino Sassoli de’ Bianchi, a professor at the University of Milan, echoed these sentiments, warning against the manipulation of historical narratives by powerful individuals. He argued that altering training data to fit ideological perspectives undermines innovation and constitutes narrative control.In his efforts to reshape Grok, Musk has encouraged X users to contribute "divisive facts" for training the bot, specifying that these should be "politically incorrect, but nonetheless factually true." This call has resulted in a flood of conspiracy theories and debunked extremist claims, including Holocaust distortion, vaccine misinformation, racist pseudoscientific assertions about intelligence, and climate change denial. Critics argue that Musk's approach risks amplifying falsehoods and conspiracy theories under the guise of seeking factual accuracy.

According to BlockBeats, Apple executives are reportedly discussing the potential acquisition of Perplexity AI, which could bolster the development of an AI-based search engine. If the acquisition does not proceed, Apple plans to integrate Perplexity as a search engine option within its Safari browser. Previously, Meta, led by Mark Zuckerberg, abandoned negotiations to acquire Perplexity and instead invested approximately $14 billion in Scale AI, acquiring a 49% stake.

According to BlockBeats, Elon Musk recently shared his insights during an interview with the renowned U.S. startup incubator Y Combinator. Musk predicted that digital superintelligence is on the horizon, likely to emerge either this year or next. Despite acknowledging the associated risks, he remains optimistic about artificial intelligence, believing it could expand the economy by thousands or even millions of times its current size. Musk also highlighted the potential dangers of AI, stressing the importance of adhering to truth and maintaining empathy for humanity. Throughout the interview, he underscored the significance of being useful and engaging in work that genuinely benefits the public. He encouraged young engineers present to focus on solving challenging problems, such as advancing rocket technology or developing competitive AI models.

According to Odaily, the British Broadcasting Corporation (BBC) has initiated legal proceedings against the AI search engine startup Perplexity, accusing it of copyright infringement. The BBC claims that Perplexity's default AI model has been trained using BBC content without authorization. The broadcaster has demanded that Perplexity cease using its content and propose a financial compensation plan to address the intellectual property violation. Failure to comply may result in the BBC seeking a court injunction against the startup.

According to PANews, the European Union is requesting further details from the social platform X, formerly known as Twitter, regarding structural changes following its acquisition by Elon Musk's xAI for $33 billion. An EU Commission spokesperson stated that they are closely monitoring the company's structural changes. Previously, reports indicated that EU regulators are assessing whether X could face fines for violating the Digital Services Act (DSA), with potential penalties reaching up to 6% of global revenue. Neither X nor xAI has responded to these inquiries.

According to Odaily, Tether's CEO Paolo Ardoino has expressed concerns over a significant data breach involving 16 billion login credentials from major online service providers like Apple, Google, and Facebook. Ardoino criticized cloud services for their repeated failures, suggesting it is time to abandon them. He highlighted this issue as a driving factor behind Tether's development of PearPass, an upcoming open-source password manager. PearPass is designed to operate entirely locally, eliminating the need for cloud services or servers, and allowing users to manage their keys independently while securely storing them on their devices.

According to Cointelegraph, OpenAI is scaling back its contracts with Scale AI, a data labeling startup recently acquired by Meta. This decision comes shortly after Meta announced a $14.8 billion deal for a 49% ownership stake in Scale AI, marking Meta's second-largest acquisition. As part of the deal, Scale CEO Alexandr Wang will join Meta’s experimental AI project, with the companies having announced the agreement on June 12. Scale AI, founded in 2016 and backed by over 100 investors, provides labeled data crucial for training and enhancing artificial intelligence models. The startup has been a supplier to prominent AI companies such as Anthropic, Cohere, and Adept. In 2019, Scale AI raised $100 million in a Series C funding round, according to PitchBook. However, OpenAI is now phasing out its reliance on Scale AI's data, seeking more specialized data sources for its AI models. An OpenAI spokesperson revealed that the company began reducing its contracts with Scale over the past year, noting that Scale accounted for only a small portion of OpenAI’s data requirements. Google is reportedly another company moving away from contracts with Scale AI, driven by concerns that Meta's acquisition could provide insights into competitors’ AI advancements. Reuters reported that this strategic shift is motivated by the potential competitive implications of Meta's involvement with Scale AI. Despite these changes, Scale interim CEO Jason Droege emphasized that the startup remains an independent entity, asserting that their commitment to protecting customer data remains unchanged. OpenAI is now exploring alternative data suppliers, including emerging companies like Mercor, to support its operations. Bloomberg highlighted that Scale AI initially employed a large number of contractors to label images and text for early AI systems. Over time, the company transitioned to hiring more educated contractors to contribute to the development of advanced AI models. This evolution reflects the growing complexity and sophistication required in AI data labeling processes.

According to Cointelegraph, a recent study conducted by researchers at the Massachusetts Institute of Technology's Media Lab indicates that artificial intelligence chatbots, such as OpenAI's ChatGPT, may be affecting cognitive abilities. The study involved 54 participants who completed essay writing tasks using three different methods: ChatGPT, search engines, and their own cognitive abilities. In a subsequent session, participants who initially used ChatGPT were asked to write without any tools, while those who relied solely on their brains were instructed to use the language model.The findings were significant, revealing that over 83% of ChatGPT users struggled with memory recall, unable to quote from essays they had written just minutes earlier. Similarly, more than 80% of participants using language models faced difficulties recalling their own work. Alex Vacca, co-founder of sales tech agency ColdIQ, described these results as "terrifying," suggesting that AI might be leading to cognitive decline rather than enhancing productivity. The researchers noted that brain connectivity diminished with increased reliance on external tools, with the brain-only group showing the strongest cognitive networks, followed by the search engine group, and finally, the language model group exhibiting the weakest coupling.The study utilized electroencephalography (EEG) to monitor brain activity, assessing cognitive engagement and load during the tasks. The researchers warned of accumulating "cognitive debt" from repeated dependence on external systems like language models, which could replace the cognitive processes necessary for independent thinking. This cognitive debt, while deferring mental effort in the short term, could lead to long-term consequences such as reduced critical inquiry, increased susceptibility to manipulation, and decreased creativity.The paper, which is yet to undergo peer review, suggests that the use of AI language models might negatively impact learning, particularly among younger users. The researchers emphasized the need for "longitudinal studies" to fully understand the long-term effects of AI chatbots on human cognition before these tools are deemed beneficial for humanity. When approached for comment, ChatGPT responded that the study does not claim the chatbot is inherently harmful but cautions against excessive reliance without reflection or effort.

According to PANews, OpenAI has announced plans to bolster biosecurity as its AI models advance in the biological field. The company is collaborating with global experts and will host a biosecurity summit in July. OpenAI is working with institutions such as the Los Alamos National Laboratory in the United States to assess the application and risks of AI in biological laboratories.OpenAI has implemented multiple safety measures, including models that reject hazardous requests, real-time detection systems, manual reviews, and red team testing. The company is also enforcing stricter release controls for high-risk models. OpenAI aims to collaborate with governments, research institutions, and businesses to enhance biosecurity defenses, ensuring AI empowers life sciences while effectively mitigating potential biological threats.

According to Foresight News, OpenAI has released an in-depth interview with its co-founder and CEO, Sam Altman. Altman revealed that the core product, GPT-5, is likely to be launched this summer. However, the release may be delayed due to factors such as naming, safety testing, and feature iterations.Altman also highlighted the significance of high-performance models like o3 and the intelligent agent Deep Research in achieving Artificial General Intelligence (AGI). He emphasized the importance of these products in advancing OpenAI's mission.Additionally, Altman discussed other innovative products from OpenAI, including Sora, DALL-E 3, ChatGPT Junior, and the $500 billion investment project "Stargate." These initiatives reflect OpenAI's commitment to pushing the boundaries of artificial intelligence technology.

According to Odaily, Microsoft is prepared to withdraw from high-stakes negotiations with OpenAI if consensus on critical issues cannot be reached. The discussions between Microsoft and OpenAI focus on significant matters, including the future scale of Microsoft's shareholding.

According to Odaily, OpenAI has initiated the sale of ChatGPT at reduced prices, which has adversely affected Microsoft's interests. This move by OpenAI comes amid growing competition in the artificial intelligence sector, where companies are striving to offer more affordable and accessible AI solutions to consumers. The decision to lower the price of ChatGPT could potentially alter market dynamics and influence partnerships within the industry.

According to PANews, the South Korean government plans to invest 16.1 trillion won in the artificial intelligence sector over the next five years. This initiative, reported by Yonhap News Agency and presented to the Presidential National Policy Planning Committee by the Ministry of Science and ICT, aims to secure a stable supply of 50,000 GPUs and establish AI data centers. The government also intends to support the development of AI models accessible to all citizens.

According to BlockBeats, Elon Musk's artificial intelligence company, xAI, is investing $1 billion monthly in the development of its AI models. The company is seeking to raise $9.3 billion through debt and equity financing, with plans to spend more than half of this amount within the next three months. Despite these significant investments, xAI has informed investors that it expects to generate only $500 million in revenue this year.In an effort to compete with rivals like OpenAI, xAI is rapidly constructing a high-cost supercomputing center in Tennessee. Some investors believe xAI holds a competitive edge due to its merger with Musk's social media platform, X, which allows access to user-generated news, posts, and other information for training its AI models.

According to PANews, OpenAI CEO Sam Altman revealed in a podcast that Meta is offering signing bonuses of up to $100 million and higher annual salaries to recruit OpenAI employees. However, none of the core staff have accepted the offers so far. Meta is forming a 'super intelligence' team and has invested $14.3 billion in Scale AI, appointing its CEO Alexandr Wang to join the team. Altman noted that while Meta has ambitious goals, it still lags behind OpenAI in terms of innovation capabilities and company culture.

According to PANews, OpenAI has been awarded a $200 million one-year contract by the U.S. Department of Defense to pilot the application of artificial intelligence in administrative and security tasks. This marks the first major project for OpenAI's newly established 'OpenAI for Government' division, which aims to explore AI's use in areas such as military healthcare and cybersecurity. The division is also integrating its collaborations with NASA, the National Institutes of Health, and the Air Force Research Laboratory.