Жаңалықтар

According to Odaily, OpenAI CEO Sam Altman hinted at the release of a new AI model, GPT-o3, through a tweet on the X platform. Altman's message included phrases like "hohoho, see you tomorrow" and "find a clue, should say oh oh oh," suggesting the potential launch of the o3 model. The repeated use of "o" and "h" in his tweet is seen as an indication of the upcoming release. OpenAI is currently developing this next-generation inference model, which aims to provide deeper insights into user queries. The company may have chosen to skip the "o2" naming to avoid trademark conflicts with the UK telecom operator O2, opting instead for "o3." The announcement coincides with the final day of OpenAI's "12 Days of OpenAI" event. Over the past 11 days, OpenAI has released a fully-featured o1 model and introduced related APIs. Additionally, the company has upgraded various features, including function calling, structured output, inference effort control, developer messaging, and visual input capabilities.

US Secretary of State Antony Blinken has emphasized the need for unified global action on artificial intelligence (AI). Speaking on the importance of regulation, Blinken stated, "We must use our collective strength to establish, update, and ultimately implement international AI standards, which are essential to lasting security."

According to Foresight News, Cointelegraph reports that the Bavarian State Office for Data Protection Supervision (BayLDA) in Germany has concluded its investigation into the digital identity project World, formerly known as Worldcoin, regarding its compliance with the European Union's General Data Protection Regulation (GDPR). The regulatory body has mandated that World implement a data deletion procedure that aligns with GDPR standards within one month of the ruling's enforcement.World has responded to this decision by filing an appeal. The company is seeking judicial clarification from the regulatory authorities on whether its privacy-enhancing technologies (PET) meet the EU's legal definition of anonymity. This appeal highlights the ongoing debate and challenges faced by digital identity projects in aligning with stringent data protection laws while maintaining their operational integrity and technological advancements.

According to PANews, OpenAI has announced that ChatGPT now supports interaction via phone calls for users in the United States. By dialing the dedicated number 1-800-CHAT-GPT (1-800-242-8478), users can engage with ChatGPT using both landlines and traditional mobile phones. This new feature is also integrated with WhatsApp, allowing users to send inquiries through the same number, offering functionality similar to the existing web and mobile versions.Currently, U.S. users are entitled to 15 minutes of free voice calls per month, although OpenAI notes that specific rules and limitations may be subject to change. The company advises users seeking a more comprehensive experience, including additional features like search capabilities, higher usage limits, and personalized interactions, to continue accessing ChatGPT directly through their accounts. This development marks a significant expansion in the accessibility of ChatGPT, aiming to enhance user interaction through diverse communication channels.

According to Cointelegraph, the United Kingdom's government has initiated a consultation with the artificial intelligence (AI) and creative industries to explore potential legal frameworks for AI models to utilize copyrighted material. The proposals, released on December 17, invite feedback from both sectors until February 25, 2025. Peter Kyle, Secretary of State for Science, Innovation and Technology, highlighted that the ambiguity surrounding copyright law's application to AI is hindering both industries from achieving their full potential. He stated that the current AI and copyright framework does not adequately support the creative and AI sectors in competing globally. This consultation arises amid criticism faced by AI companies accused of using intellectual property without permission to train their models. Among the four policy options proposed by the UK government is allowing AI companies to use copyrighted material without the right holders' consent, permitting commercial use with minimal restrictions. Another option suggests that companies can freely use copyrighted material for AI training unless creators opt out. A third proposal involves strengthening copyright laws, mandating licensing for all cases, ensuring companies only train AI models with licensed and permitted work. The consultation also considers maintaining the current laws, though this would perpetuate the existing lack of clarity for copyright holders and AI developers. The proposals have drawn criticism from creatives. Ed Newton-Rex, a British composer and CEO of Fairly Trained, a nonprofit certifying AI companies that obtain licenses for their training data, expressed concerns that the changes would primarily benefit AI companies while causing significant harm to creators. He argued that some changes are misleading, as a copyright exception would legalize training on copyrighted work without a license, which is currently illegal. Owen Meredith, chief executive of the News Media Association, also criticized the consultation in a December 17 statement, arguing that it fails to address the core issue of enforcing robust transparency requirements to protect creatives' rights. He contended that the proposals would complicate the law and allow AI firms to evade responsibilities. Meredith suggested that instead of proposing unworkable systems like 'rights reservations' or 'opt-out' regimes, the government should focus on implementing transparency requirements within the existing copyright framework.

According to PANews, Chris Dixon, founder and managing partner of a16z crypto, has highlighted the transformative impact of artificial intelligence (AI) on the current internet economy. He notes that AI is not only disrupting existing models but also reinforcing the dominance of major tech companies. This shift is accompanied by challenges such as deepfake technology and fake accounts, which undermine user trust in online platforms.Dixon suggests that blockchain technology offers a decentralized solution to these issues, empowering users to regain control over the internet. Blockchain provides a transparent and immutable mechanism that can enhance the management of digital assets and creator content ownership. It also plays a crucial role in verifying identities, thereby isolating fake accounts and imposters, and improving authenticity and security through tamper-proof content records.By integrating AI with blockchain technology, there is potential for the internet to remain open, diverse, and innovative, staying true to its original core principles. This combination could effectively address the challenges posed by AI, ensuring a balanced and secure digital environment.

According to Odaily, a special stock sale arrangement between OpenAI and Japan's SoftBank Group will allow approximately 400 current and former OpenAI employees to cash in millions of dollars. Sources familiar with the matter revealed that this tender offer permits certain OpenAI employees and former employees to sell their shares to SoftBank at $210 per share. This transaction enables employees to monetize part of their equity, with the company's valuation having doubled over the past year to $157 billion.SoftBank plans to purchase up to $1.6 billion worth of shares from eligible shareholders, who must have received restricted stock units more than two years ago to participate. Eligible OpenAI shareholders must decide by December 24 whether to sell and how much to sell. OpenAI currently employs over 2,000 people. If the offer is oversubscribed, current OpenAI employees will be prioritized over former employees. Each eligible current employee can cash out up to $10 million. Depending on participation, former employees may not be able to sell the full $10 million in equity, but according to OpenAI's rules, they will be able to cash out at least $2 million worth of stock.Sources indicate that OpenAI has stated the total amount of stock eligible for sale is $2 billion. With SoftBank agreeing to purchase $1.6 billion, this leaves potentially $400 million worth of stock without a buyer.

According to PANews, OpenAI has announced the official release of its o1 model API during the ninth day of its technology sharing live stream. This release includes significant upgrades to the real-time API, incorporating WebRTC support and reducing costs by 60%. The o1 model is equipped with advanced reasoning capabilities, supporting function calls, structured output, and visual features. Compared to its preview version, the thinking cost has been reduced by 60%. In addition to these enhancements, the cost of GPT-4o audio has decreased by 60%, and the price of the mini version has been reduced to one-tenth of its original cost. OpenAI has also introduced a new preference fine-tuning method, utilizing a direct preference optimization algorithm to better align the model with user and developer preferences. Furthermore, developers can now use newly added Go and Java SDKs, making it easier to integrate OpenAI's functionalities into their applications.

According to Odaily, Stanford University's 'Future of Digital Currency Initiative' (FDCI) has announced a groundbreaking research partnership with Eliza Labs, the creators of the open-source Eliza agent framework. This collaboration aims to explore how autonomous AI agents could transform digital currency systems. The partnership is set to commence in the first quarter of 2025, combining Stanford's expertise in digital currency research with Eliza Labs' cutting-edge capabilities in autonomous agent development. This collaboration marks the establishment of the first AI x Web3 laboratory within Stanford's 'Future of Digital Currency Initiative.' The program will focus on open-source frameworks for trust mechanisms, scalable coordination protocols, and decentralized governance models for agents. Early research findings are expected to be shared through peer-reviewed publications and industry demonstrations. This initiative represents a significant step in understanding and developing the intersection of AI and digital currency, potentially paving the way for innovative solutions in the financial technology sector.

According to Cointelegraph, the Australian government is set to develop a National AI Capability Plan aimed at enhancing the country's artificial intelligence industry. This initiative comes as Australia seeks to improve its competitiveness in the rapidly expanding AI sector. Ed Husic, the Minister for Industry and Science, announced on December 16 that the plan will explore strategies to bolster the AI ecosystem by upskilling the workforce, attracting investment, and fostering innovation.Husic emphasized the importance of leveraging AI expertise to secure supply chains and strengthen critical infrastructure. He stated that the government will collaborate closely with firms, communities, and workers to drive investment in AI capabilities. As part of the review process, the government agency will assess existing state and federal support mechanisms to determine their impact on the AI industry. Additionally, the review will focus on increasing AI literacy and enhancing the local workforce through training and re-training, while also identifying research and innovation opportunities across universities and businesses.The Australian Government Department of Industry, Science and Resources reports that the country currently hosts approximately 650 AI companies. AI and automation are projected to contribute hundreds of billions annually to Australia's GDP by 2030. The National AI Capability Plan is expected to be completed by the end of 2025, following a period of targeted and public consultation. However, some industry experts argue that the timeline is too lengthy for the fast-evolving sector, which Statista predicts could reach $826 billion globally by 2030.Simon Bush, CEO of the Australian Information Industry Association (AIIA), expressed that while the National AI Capability Plan is a positive step, the timeline should be accelerated. He suggested that the review deadline be moved to July 2025, as the current schedule could delay funding until at least 2027, which he considers too long for the rapidly advancing AI landscape. Bush noted that Australia has been slow to adopt AI across its economy compared to global standards due to concerns and fears surrounding adoption. He emphasized the need for a balanced approach that ensures AI regulation and safeguards protect citizens while also supporting AI innovation, investment, and adoption.

According to BlockBeats, on December 16, SlowMist founder Yu Jian highlighted a noticeable divergence in the development paths of AI agents within the crypto and non-crypto industries.In the crypto sector, the focus is primarily on creating incentive mechanisms through the issuance of tokens. This approach aims to leverage blockchain technology to enhance engagement and participation within the ecosystem. The crypto industry seeks to incentivize users and developers by utilising tokens, fostering a community-driven environment that encourages innovation and collaboration. This token-centric strategy is seen as a way to align the interests of various stakeholders and drive the adoption of AI technologies in decentralized applications.Conversely, traditional tech giants are making significant strides in developing interoperability protocols and practical AI applications. These companies are concentrating on creating seamless integration between different systems and platforms, enhancing the utility and accessibility of AI technologies. By focusing on interoperability, tech giants aim to break down silos and enable AI solutions to work across diverse environments, thereby increasing their applicability and effectiveness. This approach emphasizes the practical deployment of AI, ensuring that it can be utilized in a wide range of real-world scenarios, from enterprise solutions to consumer applications.The divergence in these development paths underscores the differing priorities and strategies of the crypto and tech sectors. While the crypto industry is driven by the potential of decentralized finance and community engagement, traditional tech companies are leveraging their resources and expertise to push the boundaries of AI technology and its integration into everyday life. This contrast highlights the dynamic nature of AI development and the varied approaches to harness its potential across different industries.

According to Odaily, Elon Musk's artificial intelligence company, xAI, has launched an upgraded version of its Grok chatbot. This new version introduces a Grok button on the X platform, enhancing user interaction. The updated chatbot boasts a threefold increase in speed, alongside improvements in accuracy and adherence to instructions.The enhancements in the Grok chatbot are part of xAI's ongoing efforts to refine artificial intelligence technologies. By increasing the speed and precision of the chatbot, xAI aims to provide users with a more efficient and reliable tool for communication. The addition of the Grok button on the X platform is expected to streamline user access and engagement, further integrating the chatbot into everyday digital interactions.These developments reflect xAI's commitment to advancing AI capabilities, positioning the company at the forefront of innovation in the field. The improvements in Grok's performance are anticipated to enhance user experience significantly, offering a more responsive and accurate AI assistant. As xAI continues to evolve its technologies, the company remains focused on delivering cutting-edge solutions that meet the growing demands of AI users worldwide.

According to PANews, Elon Musk has responded to a suggestion regarding the introduction of an 'X Mail' service. Musk indicated that this idea is already on his to-do list, suggesting the possibility of a future email service with the format 'username@X.com'.

According to PANews, Kimi has introduced the k1 visual thinking model, which is built on reinforcement learning technology. This model natively supports end-to-end image understanding and chain-of-thought technology, extending its capabilities beyond mathematics to various fundamental scientific fields.In benchmark capability tests across disciplines such as mathematics, physics, and chemistry, the initial k1 model has outperformed global standard models, including OpenAI's o1, GPT-4o, and Claude 3.5 Sonnet. This advancement marks a significant step in the development of models that can handle complex scientific tasks, showcasing the potential of k1 to revolutionize how these subjects are approached and understood.The k1 model's ability to excel in these areas highlights its advanced design and the effectiveness of reinforcement learning in enhancing model performance. As it continues to evolve, k1 is expected to contribute significantly to the field of artificial intelligence, offering new insights and solutions across various scientific domains.

According to Odaily, OpenAI co-founder Ilya Sutskever recently delivered a speech at the NeurIPS 2024 conference in Vancouver, Canada, where he discussed the impending conclusion of the AI pre-training era and the anticipated rise of superintelligent AI. Sutskever highlighted that the pace of advancements in computing power, driven by improved hardware, software, and machine learning algorithms, has surpassed the availability of data for training AI models. He likened data to fossil fuels, suggesting that it will eventually be depleted.Sutskever explained, "Data is not growing because we only have one internet. You could even say data is the fossil fuel of AI. It was created in a certain way, and now we are using it. We have reached peak data, and there will be no more—we must work with what we have." He emphasized that the current data resources are finite, and the industry must adapt to this limitation.Looking ahead, Sutskever predicted that the future of AI development will focus on agent AI, synthetic data, and inference-time computation. These areas are expected to drive the next phase of AI evolution, ultimately leading to the emergence of superintelligent AI. His insights suggest a shift in the AI landscape, where innovation will be necessary to overcome the challenges posed by limited data resources.

According to Odaily, on Thursday Eastern Time, tech giant Google announced the release of a new version of its Android XR operating system. This update is designed to offer a variety of natural interaction methods for headsets and smart glasses, aiming to enhance user experiences in the realms of virtual reality (VR) and augmented reality (AR).The new Android XR operating system is expected to significantly improve how users interact with VR and AR technologies. By focusing on natural interaction methods, Google aims to make these technologies more intuitive and accessible. This development is part of Google's broader strategy to advance its presence in the rapidly growing VR and AR markets, which are increasingly becoming integral to various industries, including gaming, education, and professional training.

According to ShibDaily, a two-month experiment featuring an AI Jesus at St. Peter’s Chapel in Lucerne, Switzerland, has ignited widespread discussion and criticism. The "Deus in Machina" project, launched in August, was developed by Philipp Haslbauer from the Lucerne University of Applied Sciences and Arts, alongside theologian Marco Schmid. The project utilized OpenAI’s GPT-4o model and an open-source version of Whisper technology to create an AI depiction of Jesus. Haslbauer explained that the AI video generator from Heygen was used to produce voice and video representations based on a real person. Despite concerns, no additional safeguards were implemented as GPT-4o was observed to handle controversial topics well.The AI Jesus, displayed on a computer screen within a confessional booth, engaged visitors by answering questions on faith, morality, and modern issues, providing responses rooted in Scripture and capable of communicating in over 100 languages. Visitors, primarily aged 40 to 70, included Christians, agnostics, atheists, Muslims, Buddhists, and Taoists. Researchers recorded approximately 900 conversations, noting that many visitors found the experience moving and user-friendly. Despite positive feedback, the project sparked controversy on social media. Bishop James Long of the United States Old Catholic Church criticized it as "not sacramental" and "sacrilege." Haslbauer noted some social media comments labeled the experiment "blasphemous" and "the work of the devil."Schmid emphasized that the AI Jesus was an artistic experiment meant to provoke thought about the intersection of the digital world and the divine, not to replace human interaction or sacramental confessions. He clarified that the project was not intended to conserve pastoral resources. The Catholic Church continues to explore the challenges and opportunities presented by artificial intelligence, especially following the rise of OpenAI’s ChatGPT. The experiment has highlighted the ongoing debate about the role of AI in religious contexts and its potential impact on traditional practices.

According to PANews, OpenAI's new video generation AI model, Sora, has come under scrutiny for allegedly using content from popular video games such as 'Super Mario' and 'Call of Duty,' as well as images of Twitch streamers, for its training. Legal experts have raised concerns that this could lead to multiple copyright disputes.Experts highlight that video game footage typically involves several layers of copyright, including those held by game developers, unique content created by players, and user-generated maps. Utilizing such content without authorization to train AI models could be considered infringement. Even if AI companies claim fair use, users who publish or use the generated content that is protected by copyright may still face legal risks. Furthermore, if the output involves trademarks or likeness rights, it could pose additional legal challenges.

According to Foresight News, Elon Musk recently announced that Grok, the artificial intelligence service under his leadership, has the capability to explain meme jokes.