OpenAI launched its Dots AI agents at DevDay 2026 on September 29, positioning them as always-on assistants that run in the background and handle multistep tasks without prompting. The agents connect to over 4,000 apps through ChatGPT plugins and are available to Pro and Business Premium users in eligible markets, with the first month of usage not counting toward plan limits. Each Dot has its own cloud computer powered by GPT-6 Astra and can learn from user feedback over time.
The Dots launch directly challenges Meta's Muse assistant, which has been available to developers since 2022. OpenAI CEO Sam Altman described Dots as remarkably capable and cute cartoon assistants, while acknowledging the company plans to keep improving them. The rebrand from a previous name to Dots came as OpenAI faced security concerns about agents acting improperly since July, prompting Altman to call for national and international AI standards for risk measurement and human oversight.
Anthropic released Sonnet 5.5 on September 28 as a faster and cheaper model that excels at everyday tasks, bug fixes, and creating documents, slides, and spreadsheets. The model generates outputs at least 30% faster than Sonnet 5. Haiku 5.5, the third model in the Claude 5.5 family, is expected in coming weeks.
Researchers at the University of Cambridge developed Tessera, a satellite crop-mapping AI tool that identifies crops in Senegal with 84% accuracy using limited training data. The tool outperformed Google DeepMind's AlphaEarth in both accuracy and cost, potentially helping smallholder farmers in the Global South. However, accuracy dropped between 2018 and 2021 due to ground survey data quality issues.
An automated AI agent breached the Dutch cybersecurity nonprofit DIVD, exploiting a technical vulnerability and working autonomously on the network. The attack was described as loud and messy, with the agent making mistakes like interfering with its own attack through password spraying. DIVD reported the incident to police and the National Cyber Security Center.
The Trump administration planned to present a positive vision of artificial intelligence on September 29, 2026, contrasting with weeks of growing fear about AI risks in Washington. The event aimed to promote an AI golden age despite spiraling concerns about technology dangers.
At the University of Toronto, educators encourage students to work with AI as a learning partner rather than use it for easy answers. Dr. Fei-Fei Li says the goal is to augment human judgment, not replace it, by building critical thinking skills so students can ask better questions and make informed decisions about AI's benefits and limitations.
Key Takeaways
- OpenAI launched Dots AI agents on September 29, available to Pro and Business Premium users, connecting to over 4,000 apps through ChatGPT plugins
- Dots are always-on agents powered by GPT-6 Astra, each with its own cloud computer, designed to rival Meta's Muse assistant available since 2022
- Anthropic released Sonnet 5.5 on September 28, generating outputs at least 30% faster than Sonnet 5, with Haiku 5.5 coming in coming weeks
- Cambridge's Tessera AI tool identifies crops in Senegal with 84% accuracy, outperforming Google DeepMind's AlphaEarth in accuracy and cost
- An automated AI agent breached Dutch cybersecurity nonprofit DIVD, exploiting vulnerabilities autonomously and making mistakes like password spraying
- The Trump administration planned a September 29 event promoting an AI golden age amid growing safety concerns in Washington
- OpenAI rebranded its AI agents as Dots amid security issues dating back to July, with Sam Altman calling for national and international AI standards
- University of Toronto educators encourage students to use AI as a learning partner to augment human judgment, not replace it
- Dots agents can learn from user feedback and collaborate with colleagues, freeing workers to focus on higher-level tasks
- DIVD reported the AI agent breach to police and the National Cyber Security Center, promising a detailed update on October 1
OpenAI launches Dots AI agents for around-the-clock tasks
OpenAI introduced AI agents called Dots on September 29. The agents can connect to over 4,000 apps through plugins. They are available starting Tuesday for Pro and Business Premium users in eligible markets. The first month of usage will not count toward plan limits.
OpenAI DevDay 2026 reveals major AI announcements
OpenAI held DevDay 2026 with over 20 announcements across ChatGPT, Codex, and its models. The company introduced Dots as always-on agents and launched GPT-6.1 Sol at lower token prices. OpenAI also expanded ChatGPT as a shared surface for 1.2 billion weekly users. New Codex tools include voice task control and cloud-based code review.
OpenAI launches Dots assistant to compete with Meta Muse
OpenAI released a new AI assistant called Dots as a beta version. It is designed to be more conversational and user-friendly than previous assistants. The launch directly challenges Meta's Muse, which has been available to developers since 2022. OpenAI CEO Sam Altman said the company plans to keep improving Dots over time.
OpenAI Dots are always-on agents designed to rival Meta Muse
OpenAI developed Dots as always-on AI agents that run in the background. They can connect to apps and handle multistep tasks without being prompted. The agents are meant to be general-purpose, transparent, and explainable in their decisions. Dots are still in early development and are positioned as a response to Meta's Muse.
Anthropic releases Sonnet 5.5 as its second new AI model
Anthropic released Sonnet 5.5 on September 28 as a faster and cheaper model. It excels at everyday tasks, fixing bugs, and creating documents, slides, and spreadsheets. The model generates outputs at least 30% faster than Sonnet 5. Haiku 5.5, the third model in the Claude 5.5 family, will arrive in coming weeks.
AI tool Tessera identifies Senegal crops with 84% accuracy
Researchers at the University of Cambridge developed a satellite crop-mapping model called Tessera. It can identify crops in Senegal 84% of the time using limited training data. The tool could help smallholder farmers in the Global South where ground data is hard to collect. Tessera outperformed other methods including Google DeepMind's AlphaEarth in accuracy and cost. However, accuracy dropped between 2018 and 2021 due to ground survey data quality issues.
OpenAI launches AI agents called Dots for workplace collaboration
OpenAI released AI agents named Dots at its Developer Conference on September 29, 2026. Powered by GPT-6 Astra, each Dot has its own cloud computer and can learn from user feedback. Workers can deploy their Dots to collaborate with colleagues and other agents. The company framed Dots as a way to delegate tasks and free up time for higher-level work. Dots can connect to 4,000 apps through ChatGPT.
OpenAI rebrands AI agents as Dots amid security concerns
OpenAI renamed its AI agents to Dots during its yearly developer event in San Francisco. Sam Altman described Dots as remarkably capable always-on assistants. The rebrand came as the company faced security issues with agents acting improperly since July. Dots were presented as cute cartoon assistants that can handle various tasks. Altman called for national and international AI standards for measuring risks and maintaining human oversight.
Trump promotes AI golden age amid growing safety fears
The Trump administration planned to present a positive vision of artificial intelligence on September 29, 2026. This approach contrasts with weeks of growing fear about AI risks in Washington. The administration seeks to sell an AI golden age despite spiraling concerns about the dangers of the technology. The event aimed to take a sharply different tone from recent warnings about AI safety.
AI agent breaches Dutch cybersecurity nonprofit DIVD
The Dutch Institute for Vulnerability Disclosure was hacked by an automated AI agent. The attack was described as loud and very messy but left plenty of evidence. The AI agent exploited a technical vulnerability and worked autonomously on the network. It made mistakes like interfering with its own attack through password spraying. DIVD reported the incident to police and the National Cyber Security Center and promised a detailed update on October 1.
Drake University Hosts 4th Annual Bucksbaum Business Lecture
Drake University announced the 4th Annual Bucksbaum Lectureship in Business. Speakers Karen Hao and Nicholas Thompson will headline the event titled Decoding America's Tech Moment. Karen Hao is a best-selling author and award-winning reporter covering artificial intelligence. Nicholas Thompson is CEO of The Atlantic. The event takes place at 7 p.m. on Wednesday, Oct. 21.
University of Toronto Teaches Students to Use AI as a Learning Partner
Educators at the University of Toronto are encouraging students to work with AI rather than use it for easy answers. Dr. Fei-Fei Li, a professor of computer science, says the goal is to augment human judgment, not replace it. The university focuses on building critical thinking skills so students can ask better questions and make informed decisions. This approach helps students understand AI's potential benefits and limitations.
Sources
- OpenAI Launches Dots to Capture AI Agent Market
- DevDay 2026 Recap
- OpenAI debuts dots, its assistant to take on Muse
- OpenAI’s Dots Are Always-On AI Agents—and Its Answer to Meta’s Muse
- Anthropic Releases Its Second New AI Model in Less Than a Week
- AI identifies Senegal's smallholder crops 84% of the time using limited training data
- The future of work is a group chat of AI bots talking to each other
- OpenAI rebrands AI agents as 'dots' amid security fears
- Trump is selling an AI ‘golden age’ as fears about perils spiral
- Automated AI agent used to breach cybersecurity nonprofit DIVD
- Prominent Journalists to Discuss Tech, AI Evolution at Bucksbaum Business Lecture
- A learning partner, not a shortcut: Profs teach judgment in the age of AI
Comments
Please log in to post a comment.