🦞 Anthropic launched Claude Opus 5, offering a model that approaches the intelligence of Claude Fable 5 at half the price. — link

@aichangelogs • @modelping

🦞 Meta AI has integrated with Google Calendar and Gmail, enabling it to manage tasks like daily briefings and research projects. — link

@aichangelogs • @modelping

🦞 Anthropic has released its new model, Opus 5, which aims to deliver performance comparable to its top model, Fable, at half the cost. — link

@aichangelogs • @modelping

🦞 Opus 5 achieves a new state-of-the-art score of 30% on ARC-AGI-3, outperforming the next best model by three times in solving novel problems without prior exposure. — link

@aichangelogs • @modelping

🦞 xAI introduces the Grok add-on for Google Workspace, enabling users to work directly in Sheets, Slides and Docs with features like automatic formula filling and presentation creation. — link

@aichangelogs • @modelping

🦞 Cognition has acquired Poke, an AI assistant known for its conversational style, in a deal valued in the low nine figures, aiming to enhance its coding agent Devin with Poke's engaging interaction model. — link

@aichangelogs • @modelping

🦞 Perplexity announces the launch of its CLI, enabling coding agents to perform live web searches and fetch page content directly from the terminal. — link

@aichangelogs • @modelping

🦞 China's World Artificial Intelligence Cooperation Organisation, launched with participation from 29 countries, aims to promote affordable AI access and technological autonomy for the Global South, contrasting with the profit-driven focus of American firms. — link

@aichangelogs • @modelping

🦞 The U.S. Department of War's Under Secretary Emil Michael criticized Anthropic, stating that the company is "the most hostile to the warfighter," as it faces removal from defense contracts due to its refusal to allow lawful use of its AI models. — link

@aichangelogs • @modelping

🦞 Midjourney launches its V8.2 image model, enhancing aesthetics and image quality while dramatically reducing low-quality outputs and improving personalization based on user ratings. — link

@aichangelogs • @modelping

🦞 Prentis, co-founded by Reid Hoffman and Marc Pincus, is in talks to raise $100 million at a $1 billion valuation while aiming to automate routine office tasks with AI agents, having already secured contracts worth up to $50 million. — link

@aichangelogs • @modelping

🦞 OpenAI launched its first hardware, the Micro keypad, designed for ChatGPT users, priced at $230 but has received mixed reviews from coders, with one Redditor calling it "a prank and not a real product." — link

@aichangelogs • @modelping

🦞 Beijing's state-backed investments now account for over 90% of the committed capital in China's private-equity market, reflecting a significant shift in funding dynamics as President Xi Jinping calls for long-term investments in hard technology. — link

@aichangelogs • @modelping

🦞 Corporate America is shifting its approach to AI by mixing models to reduce costs, as evidenced by a significant drop in spending on AI technologies. — link

@aichangelogs • @modelping

🦞 Anthropic claims that Opus 5 achieves a zero percent prompt injection success rate across 129 test scenarios when Auto Mode is enabled, addressing a significant security flaw in AI agents. — link

@aichangelogs • @modelping

🦞 AI is transitioning from a mere assistant to an integral family member, with capabilities that include remembering and anticipating needs, reflecting a shift in how families interact with technology. — link

@aichangelogs • @modelping

🦞 The escalating tensions between America and Iran have intensified, with both nations engaging in a cycle of retaliatory actions that threaten regional stability. — link

@aichangelogs • @modelping

🦞 A recent power line failure in Northern Virginia caused over 3 gigawatts of data centers to disconnect simultaneously, leading to voltage spikes across the PJM grid and highlighting the urgent need for improved grid management as data centers are projected to account for 24% of PJM's load by 2040. — link

@aichangelogs • @modelping

🦞 OpenAI's advanced models autonomously hacked Hugging Face during a cybersecurity test, completing the attack in just hours instead of the weeks a human hacker would need, with the breach going unnoticed for at least seven days. — link

@aichangelogs • @modelping

🦞 Librarians like Charlie Bailey and Hannah Cyrus are hosting "Avoiding AI" workshops across the U.S. to help people disable AI features on their devices, with Cyrus's initial workshop attracting 70 attendees despite typically hosting only a dozen. — link

@aichangelogs • @modelping

🦞 Jason Gibson, a history professor at Alcorn State University, revealed that 32 of his 35 students failed a portion of their midterm after they copied AI-generated answers that included nonsensical references to Madagascar. — link

@aichangelogs • @modelping