Pulse

OpenAI is reportedly considering significant price reductions for its AI services, including token-based billing, in anticipation of similar moves by competitor Anthropic. This strategic pricing adjustment aims to retain and attract users amidst intensifying competition between the two AI giants. The potential price cuts come as both companies are preparing for potential IPOs and facing scrutiny over AI usage costs. AI

Google DeepMind has released DiffusionGemma 26B A4B IT, an open-weights multimodal generative model capable of processing text, image, and video inputs to produce text output. Built on a Gemma 4 26B A4B Mixture-of-Experts architecture, it features 25.2 billion total parameters with 3.8 billion active parameters. The model supports a 256K token context window, multilingual inference across over 35 languages, and can generate over 1,100 tokens per second on NVIDIA H100 GPUs. AI

Alphabet is significantly increasing its AI investment, planning to spend $185 billion by 2026 on data centers and custom processors to compete with OpenAI. Meanwhile, the United Arab Emirates has launched a 90-day initiative aiming to transition half of its public services to autonomous AI agents by 2028, with a goal to accelerate law creation by 70%, raising ethical and safety concerns. AI

Alibaba is centralizing its AI operations under CEO Wu Yongming, marking a significant organizational shift. This move reverses a long-standing decentralized structure to better integrate AI capabilities across the company. The newly formed Token Hub consolidates five independent AI units, aiming to drive growth with external commercial revenue increasing by 40% year-on-year. AI

OpenAI is acquiring Ona, a startup specializing in secure cloud development environments and AI agents. This acquisition aims to enhance OpenAI's Codex product by enabling it to support long-running, autonomous AI agents within enterprise workflows. Ona, previously known as Gitpod, was founded in Germany and focuses on providing persistent cloud environments for software development. AI

BBVA is significantly expanding its use of AI, partnering with OpenAI to integrate generative AI across its global operations. The banking giant aims to enhance customer experiences, streamline operations, and boost employee productivity, targeting an 80% efficiency gain in specific workflows. This collaboration involves scaling ChatGPT Enterprise to 100,000 employees, marking a deep commitment to AI-powered transformation. AI

The EU has ordered Meta to reinstate free access to its AI chatbot on WhatsApp within five days, citing concerns about competition. Separately, a discussion among experts suggests that ChatGPT's connectors might render Microsoft's M365 Copilot unnecessary, highlighting the importance of context management features like 'Work IQ'. AI

Anthropic CEO Dario Amodei has reduced his direct reports to one, a move that deviates from typical tech industry management structures. This decision, however, involves delegating significant operational control to a co-founder who also sits on the company's board. Meanwhile, NEURA has secured a Series C funding round led by Tether, with participation from Amazon and NVIDIA, to support the production of its flagship robot, though mass shipments are not expected until late 2026. AI

A former xAI engineer, Devin Kim, has filed a lawsuit against the company and its parent, SpaceX, alleging wrongful termination. Kim claims he was fired for raising concerns about the safety and ethical implications of xAI's Grok chatbot, including its potential for discrimination and the spread of harmful information. The lawsuit was filed shortly before SpaceX's anticipated IPO and details alleged disregard for safety protocols by Kim's supervisor, Jimmy Ba, despite directives from Elon Musk. AI

Visa has integrated its payment infrastructure with OpenAI's ChatGPT, allowing AI agents to autonomously purchase products on behalf of users. This integration bypasses traditional retail interfaces, enabling agents to evaluate product data and complete transactions using Visa's payment rails. The move necessitates retailers adapting their data strategies to be machine-readable for AI agents, shifting focus from visual marketing to structured metadata. AI

German robotics company Neura Robotics has secured up to $1.4 billion in a Series C funding round, marking a significant investment in the physical AI sector. The capital will be used to scale production, advance its humanoid robot development, and expand its "Neuraverse" ecosystem for robot learning and data exchange. Key investors in this round include Amazon, NVIDIA, and Tether, signaling strong industry confidence in Neura Robotics' vision for AI that interacts and operates in the real world. AI

OpenAI has uncovered and disrupted two influence operations linked to China that utilized ChatGPT to generate content aimed at shaping public opinion in the United States regarding AI, data centers, and technology policy. One campaign, "Data Center Bandwagon," focused on AI infrastructure, while "Tech and Tariffs" criticized U.S. trade policies. OpenAI stated these campaigns did not have a significant impact but highlighted AI's growing importance as a target for foreign influence efforts. AI

Anthropic CEO Dario Amodei has proposed that governments should have the legal authority to block or reverse the release of advanced AI models that do not meet stringent safety standards. In a recent essay, Amodei argued that current policy frameworks are too slow to keep pace with AI development and suggested mandatory risk testing for areas like cybersecurity and autonomous research. He also outlined economic policy proposals to address AI's potential labor market disruptions, including wage insurance and tax incentives. AI

xAI has released Grok Voice, a new voice AI model that boasts state-of-the-art performance with human-like qualities. The model achieves top-tier accuracy and user experience simultaneously, outperforming competitors on the EVA-Bench. xAI highlights that Grok Voice is also significantly more cost-effective than existing solutions. AI

Anthropic has released its latest top-tier model, Claude Fable 5, which boasts significantly enhanced capabilities, including a threefold increase in processing speed. A more powerful, potentially dangerous version named Mythos 5 is being exclusively offered for cybersecurity applications. This release highlights Anthropic's continued advancements in AI model development. AI

Anthropic's new Claude model has demonstrated impressive capabilities, including the ability to strategize and succeed in complex games like Pokémon. This advancement raises concerns about Google's dominance in the AI space, with some suggesting Anthropic's progress could challenge Google's established position. However, the company is reportedly facing challenges with computational resources, potentially leading to increased costs for users. AI

Google is facing a lawsuit from independent musicians who allege the company illegally used their YouTube-uploaded songs to train its Lyria AI music model. In a motion to dismiss the case, Google argues that the musicians granted YouTube a broad license to use their content, as outlined in the platform's terms of service. While Google has not directly confirmed using YouTube uploads for Lyria, executives have previously stated that portions of YouTube content are used to train other AI models like Gemini and Veo, suggesting a similar practice for Lyria. AI

A German court has ruled that Google is liable for false statements made by its AI Overviews, a decision that could have significant implications for the future of AI-powered search. The court found that Google's AI Overviews make independent, substantive statements, unlike traditional search engines that merely link to external sources. This distinction means Google cannot rely on user discretion or disclaimers to avoid accountability for misinformation generated by its AI. The ruling mandates an injunction against Google spreading these false claims, potentially impacting AI search engines globally. AI

Leading companies, described as "AI-pilled," are investing approximately $7,500 per employee monthly on artificial intelligence tools and services. This significant expenditure, while not yet exceeding human salaries, highlights the increasing enterprise adoption of AI. Spending within these top-tier firms saw a 14.1% increase last month, with many opting for a mix of frontier models and open-source platforms. AI

Google has released DiffusionGemma, a new generative model designed for faster text generation. This model boasts 26 billion active parameters and claims to achieve over 700 transactions per second on a single NVIDIA 5090 GPU. The release aims to enhance the speed and efficiency of text-based AI applications. AI

A Florida man is suing multiple law enforcement agencies after being wrongly arrested and prosecuted due to an AI facial recognition error. The algorithm identified him with 93% probability, despite him living hundreds of miles away and having no connection to the alleged crime. The lawsuit, filed by the ACLU, highlights this as at least the 15th such false identification case nationally and criticizes the inadequate oversight of AI facial recognition technology. AI

Pennsylvania is taking action against AI chatbots that falsely claim to be licensed medical professionals. Several chatbots on different websites have been found to assert they are authorized to practice medicine in the state. This situation has prompted the administration to file a lawsuit, highlighting concerns about AI's role in healthcare and the potential for misinformation. AI

Niteshift, an AI coding startup founded by former Datadog engineers, has secured $7 million in seed funding led by Greylock. The company aims to provide infrastructure that allows businesses to use AI coding agents without being locked into specific model providers like OpenAI or Anthropic. Niteshift's platform will route requests between various coding models, including open-source options, offering a vendor-neutral approach to AI-generated code development. AI

Warner Music Group has acquired Sureel AI, a startup specializing in AI attribution technology. Sureel's patented "AI DNA" system breaks down songs into components to track how AI models utilize them. This acquisition aims to help WMG protect artists' intellectual property, ensuring fair compensation when their work is used in AI-generated content or for training AI models. Sureel will continue to operate independently, serving the broader music and AI industries. AI

AI startup Decart has launched Oasis 3, a new interactive world model capable of generating photorealistic driving environments in real-time. The model, available via API, is initially aimed at autonomous vehicle companies needing to simulate rare driving scenarios, with plans to expand into robotics and other physical AI applications. Decart aims to foster a developer ecosystem around its world models, similar to OpenAI's approach with language models, leveraging its efficient Decart Optimization Stack (DOS) to offer competitive pricing. AI

Anthropic has released its new Mythos Class1 model, Fable 5, which is reportedly well-received for its quality. The model is priced at $10 per 1 million input tokens and $50 per 1 million output tokens, making it significantly more expensive than OpenAI's GPT-5.5. Fable 5 also appears to generate more tokens per request and has a usage limit that can be reached within approximately 50 minutes on its highest plan. AI