OpenAI unveils advanced ChatGPT agent for seamless task automation
Unlike the regular version of ChatGPT that we use currently, which only responds to queries and engages in conversation, this new agent can actively interact with websites and connected apps in real time using a 'virtual computer' environment. It can mimic human actions such as browsing the web, filling out forms, opening links and typing to complete tasks independently.
For example, if you want to plan a trip, you just need to enter your preferences and it will do everything. It can suggest places to visit, book a hotel, help with packing for the trip, provide weather information during your planned visit and much more, all without the need to enter multiple queries.
ChatGPT already has experimental tools such as Operator and Deep Research. The operator can navigate websites, while Deep Research automates complex information gathering. The new Agent brings the strengths of both of these tools together for seamless task execution and sophisticated reasoning. Additionally, users can connect apps like Gmail and GitHub, allowing the agent to scan emails, access documents,or review code repositories to enhance productivity.
OpenAI CEO Sam Altman has highlighted that while the agent is currently in the early preview phase, it has the potential to significantly boost both personal and workplace productivity by taking over repetitive and complex workflows. Importantly, users have full control over the agent and can give permission, interrupt or stop any ongoing task at any time.
This launch places OpenAI alongside other leading tech giants investing in AI agents as the future of digital assistants. The arrival of this advanced agent also suggests that the rumoured AI-powered browser could be real and possibly launching soon. There have been reports that OpenAI is working on a browser called Aura, and perhaps this new agent will power that browser.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Indian Express
28 minutes ago
- Indian Express
What is Comet, Perplexity's agentic AI-powered browser?
Perplexity, known for its AI-powered search engine, earlier this month introduced its new AI-first browser, Comet, which is designed to help users assign various web tasks to AI agents. The browser is still in early access under the Perplexity Max plan, which costs $200 a month. From user responses and early reviews, Comet seems to be offering a glimpse into the future where web browsing will be a completely different experience. It will be like users delegating a task to an agent who would browse for them to accomplish it. In essence, a user will not be directly interacting with websites, but an AI agent will be doing it in the browser for them. How is Comet different from just using the Perplexity browser? Why does the Aravind Srinivas-led AI company need to build an entirely new web browser? What can Comet do? Comet looks a lot like Google Chrome, and there is a good reason for that. The browser is built on the Chromium framework, which is an open-source architecture framework for web browsers maintained by Google. This framework is also the backbone of popular browsers such as Chrome, Microsoft Edge, and Opera. Comet would support Chrome extensions, bookmarks, and even sync with a user's settings if they import them. The AI-first browser is built from the ground up with Perplexity AI agents that are integrated directly into the browser. It not only helps users with search but also acts on their behalf across tabs, services, and platforms — this is why Perplexity describes Comet as an 'agentic browser'. Its search capabilities come from Perplexity's answer engine that is based on a mixture of foundational LLMs like GPT-4o and Claude 4.0 Sonnet, along with its proprietary model Sonar. Perplexity AI was launched a month after OpenAI's ChatGPT created ripples across the world. The AI-powered search engine rapidly grew in popularity, which stems from its unique positioning as an 'answer engine', which is radically different from a traditional search engine or AI chatbots. Perplexity was among the first to combine AI with traditional web search with accurate and real-time answers to any question. However, Comet is local and standalone in what seems like platform independence. In case Perplexity chose to build on top of Safari or Chrome, then it would likely be second to Google or Apple's native AI tools. As an independent browser, Comet gives Perplexity full control. When it comes to local context access, cloud-based browser agents would require one to log in and start from scratch. But, with Comet, the agent already knows what tab a user is viewing and what they are doing. This eliminates the need for copying, pasting, or reloading. Since it's local, the agent can interact with open pages and logged-in services instantly without additional authentication. The futuristic browser comes with some powerful features that make web browsing efficient and more productive. With Comet, you browse less and prompt more, and the AI takes care of the rest. It comes with a plethora of use cases that could likely make it a daily essential, much like what Google Chrome is to millions of users. From scheduling meetings from Gmail to comparing products across tabs, Comet is more than a browser. This is possible because it can read an email, find a suitable time for a user's calendar, draft a reply and create a tentative event with a Google Meet link. And all of these from within Gmail, without switching tabs. Similarly, if a user is shopping for a camera online, they can ask Comet to summarise and compare specs like recording quality or frame rates. If they have multiple tabs open, Comet will pull context from all these open tabs and offer a clean, structured summary to help them make an informed decision. One of the most useful use cases is the ability to summarise videos or articles. Comet can reportedly summarise an open YouTube video or web article in seconds. The video summaries are generated using transcripts and content on web pages. Some users with early access have also revealed that Comet can also accept LinkedIn requests. It can review pending requests and bulk-accept the ones it deems most relevant, saving users time. Comet's built-in assistant is placed at the top-right corner. The assistant can help in summarising the page a user is on; perform actions like clicking, filling forms or navigating; look up things across open tabs, and act simultaneously across sites. Users can even @mention tabs by name for specific tasks at open pages. For example, @YouTube – find the top comment. It needs to be noted that Comet will ask a user for confirmation before posting and notify them of the outcome. However, in some cases limitations imposed by platforms may prevent Comet from posting or engaging with comments. After years of stagnant innovation in web browsing, Comet seems to be signalling a huge shift in how the world will use the internet. Users will no longer be required to go on manually clicking through dozens of tabs, as they can now delegate tasks to agents that will understand context, objectives, and user preferences. Web browsing backed by AI agents is crucial in times when there is an overload of AI-generated content. Browsers like Comet could help filter through the clutter and deliver accurate information faster. The biggest catch with Comet for now is its availability, as it is a part of Perplexity Max, which is priced at $200 a month. Over time the company may lower the barrier or release a limited free tier if rivals OpenAI or Google move quickly with their own agentic browsers. Perplexity will roll out invite-only access slowly to its waitlist over the summer, and new users will receive a limited number of invites to share. Bijin Jose, an Assistant Editor at Indian Express Online in New Delhi, is a technology journalist with a portfolio spanning various prestigious publications. Starting as a citizen journalist with The Times of India in 2013, he transitioned through roles at India Today Digital and The Economic Times, before finding his niche at The Indian Express. With a BA in English from Maharaja Sayajirao University, Vadodara, and an MA in English Literature, Bijin's expertise extends from crime reporting to cultural features. With a keen interest in closely covering developments in artificial intelligence, Bijin provides nuanced perspectives on its implications for society and beyond. ... Read More
Time of India
31 minutes ago
- Time of India
OpenAI, Oracle deepen AI data center push with 4.5 gigawatt Stargate expansion
Synopsis OpenAI announced that Oracle will provide 2 million chips to help scale its AI data center infrastructure. This partnership aims to boost computing power for OpenAI's advanced models like ChatGPT, reflecting growing demand for AI capabilities and the need for massive processing resources to support training and deployment.
Time of India
41 minutes ago
- Time of India
1.8 billion Gmail users at risk: Experts warn of hidden threat stealing passwords silently
Google has reportedly issued a warning to 1.8 billion Gmail users around the world about a new type of scam. This sophisticated online scam uses invisible email prompts to trick its own AI assistant, Gemini into stealing passwords. According to a report by The Sun, the warning is for a specific kind of threat which is designed to fool users and lead them to reveal their login credentials. This alert also highlights the persistent threat of sophisticated cyberattacks that mainly target personal accounts online. The report adds that cybercriminals are embedding hidden instructions in emails with the help of white text and zero font size. This text is invisible to the user but can be easily read by Gemini. Whenever a user clicks on 'summarise this email' option, Gemini may generate fake security alerts and prompt the user to share their sensitive information of make calls to fraudulent support numbers. How the scam works As per the report by The Sun, the hackers embed som indirect prompt injections into emails. The Google chatbot— Gemini, then read these hidden commands and display false warnings on the screen of the user. The users are then asked to click on malicious links or call some fake support lines. The AI cannot distinguish between user queries and embedded hacker prompts which leads to the user being scammed. What Google and experts recommend Cybersecurity experts are urging all Gmail users to remain vigilant and adopt robust security practices. The experts have urged the users to not trust Gemini summaries which claim that their account has been compromised. The experts also advise users to configure email clients to detect and neutralise hidden content. The users can also use post-processing filters which will help in scanning suspicious keywords, URL's or phone numbers. Also, consider switching to passkeys for stronger, phishing-resistant authentication. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Live Update: The Strategy Uses By Successful Intraday Trader TradeWise Learn More Undo Mozilla's 0Din security team first uncovered the exploit, showing how Gemini could be manipulated into displaying a fake alert that a user's password had been stolen. Google has acknowledged the issue but has yet to fully patch the vulnerability. AI Masterclass for Students. Upskill Young Ones Today!– Join Now
