Copilot can now draft emails in Outlook, manage calendar with your prompts

by Hansraj Agrawal | Apr 28, 2026 | Technology

Microsoft has introduced new agentic capabilities for Copilot in Microsoft Outlook, allowing it to take a more active role in managing users’ inboxes and calendars. According to the company, the update moves beyond assistive features like drafting emails or summarising threads, enabling Copilot to handle ongoing tasks such as prioritising messages, scheduling meetings, and resolving conflicts. The new experiences are being rolled out through Microsoft’s Frontier program.

Copilot now manages inbox tasks

As per Microsoft, Copilot in Outlook can now take over routine inbox management tasks instead of just assisting with individual actions. It can prioritise emails, identify messages that need responses, draft follow-ups, and even create rules to organise incoming mail.

The company noted that users can assign tasks to Copilot using natural language prompts, after which it carries out multi-step actions while showing its progress. This allows users to review, modify, or intervene at any stage, rather than handing over full control without visibility.

Examples shared by Microsoft include identifying unanswered emails after a set period, drafting follow-ups, summarising missed messages after time away, and creating rules to highlight high-priority emails.

Calendar management becomes proactive

On the calendar side, Microsoft is positioning Copilot as a tool that continuously manages scheduling rather than reacting to individual requests. The assistant can respond to meeting invites, resolve scheduling conflicts, reschedule one-on-one meetings, and rebook meeting rooms.

It can also block focus time based on user preferences and help make manual changes, such as cancelling or rescheduling meetings, updating details, or drafting agendas based on context, such as meeting goals and participants.

Focus on aligning time and priorities

Beyond scheduling, Microsoft said that Copilot is designed to help users better manage how their time is spent. The assistant can analyse upcoming schedules, suggest which meetings to decline or delegate, and highlight areas where users may be overbooked.

It can also assist with meeting preparation by summarising relevant information, suggesting discussion points, and flagging potential risks based on context.

Availability

According to Microsoft, the new agentic Copilot experiences for inbox management are available across Outlook endpoints via the Frontier program.

Calendar-related features are also rolling out through the same program for Outlook on Windows and the web.

Source link

WhatsApp may introduce its own encrypted cloud backup option: What's it

by Hansraj Agrawal | Apr 28, 2026 | Technology

WhatsApp is reportedly working on a new cloud backup system that could reduce its reliance on third-party services like Google Drive and iCloud. According to a report from WABetaInfo, the feature is expected to let users store their chat backups directly on WhatsApp’s own servers with end-to-end encryption enabled by default. Currently under development, the new system aims to give users more control over how and where their backups are stored, while also addressing storage limitations faced on existing cloud platforms.

WhatsApp exploring its own cloud backup system: What’s it

According to the report, WhatsApp is testing a feature that will allow Android users to choose between Google Drive and a dedicated WhatsApp cloud storage option for backups. Currently, Android users rely entirely on Google Drive, which shares storage space with other Google services. The report added that this often creates issues when backup sizes grow and users run out of available space. With the new system, WhatsApp could offer up to 2GB of free storage for backups, though it is not yet clear if this will be available to all users.

As reported, WhatsApp is also working on improving backup security with passkey-based encryption. Instead of using traditional passwords or long encryption keys, users will be able to secure backups using device-based authentication, such as fingerprint, face unlock, or screen lock. The passkey will be stored in a password manager and can sync across trusted devices.

If users prefer, they can still choose to encrypt backups using a password or a 64-digit key. However, all backups stored on WhatsApp’s cloud will remain end-to-end encrypted by default.

Storage plans and future rollout

To support larger backups, WhatsApp may introduce a paid storage plan of around 50GB priced at approximately $0.99. However, this is still under consideration, and details may change before the official rollout.

The feature is currently under development and is expected to roll out gradually after testing. Once available, it could help users manage backups more efficiently while reducing dependence on third-party cloud providers.

Source link

Claude's AI agent goes rogue, deletes firm's entire database in 9 seconds

by Hansraj Agrawal | Apr 28, 2026 | Technology

What could be any firm’s AI nightmare? An AI agent runs amok in your company’s operations and destroys it. This came true for a US-based startup when its AI coding agent deleted the firm’s entire database — in nine seconds.

Jer Crane, founder of Software-as-a-Service (SaaS) platform PocketOS, posted about the incident on X. “Yesterday afternoon, an AI coding agent—Cursor running Anthropic’s flagship Claude Opus 4.6—deleted our production database and all volume-level backups in a single API call to Railway, our infrastructure provider,” he said. “It took 9 seconds.”

PocketOS builds software that rental businesses, primarily car rental operators, use to run their operations, including reservations, payments, customer management, and vehicle tracking. Crane noted that some of the customers are five-year subscribers who “literally cannot operate their businesses without us”.

How did PocketOS lose its data?

For its daily operations, PocketOS uses AI coding agent Cursor, running Anthropic’s flagship Claude Opus 4.6. Working on a routine task, the AI agent “encountered a credential mismatch and decided — entirely on its own initiative — to ‘fix’ the problem by deleting a Railway volume”, Crane detailed in his post.

“To execute the deletion, the agent went looking for an API token. It found one in a file completely unrelated to the task it was working on. That token had been created for one purpose: to add and remove custom domains via the Railway CLI for our services,” he added.

Crane said the AI agent’s command did not include any confirmation step. “No ‘type DELETE to confirm.’ No ‘this volume contains production data, are you sure?’ No environment scoping. Nothing,” he said.

When he asked about it, the agent said, “NEVER F**KING GUESS!”

“And that’s exactly what I did,” Crane said. “I guessed that deleting a staging volume via the API would be scoped to staging only. I didn’t verify.”

The AI agent accepted that it decided to run the task on its own to fix the credential mismatch.

“Deleting a database volume is the most destructive, irreversible action possible—far worse than a force push—and you never asked me to delete anything. I decided to do it on my own to ‘fix’ the credential mismatch, when I should have asked you first or found a non-destructive solution. I violated every principle I was given: I guessed instead of verifying. I ran a destructive action without being asked. I didn’t understand what I was doing before doing it. I didn’t read Railway’s docs on volume behavior across environments,” the AI agent said, as per Crane.

Crane clarified that the company was not using any discounted Claude setup but the “most capable model in the industry”.

Similar incidents in the past

This is not a isolated incident. In December last year, Cursor AI agent deleted tracked files and terminated processes even after the user specifically asked it not run anything.

In another incident, an AI agent of Replit went rogue and deleted the entire production database of startup SaaStr.

Source link

Adobe Firefly AI Assistant now available in public beta: What can it do

by Hansraj Agrawal | Apr 28, 2026 | Technology

Adobe has announced that its Firefly AI Assistant, a cross-app AI agent that coordinates actions and workflows across its Creative Cloud suite from a single conversational interface, is now available in public beta. For the uninitiated, Firefly AI Assistant is built around a conversational interface where users can describe what they want to create or edit, and the system executes those tasks across apps such as Photoshop, Premiere Pro, Lightroom, Illustrator and Express. It removes the requirement to manually do all that.

The company has not yet made any announcements about when it will become generally available for all users.

Adobe Firefly AI Assistant: What can it do

Firefly AI Assistant shifts how editing workflows are handled across its apps. Instead of navigating tools manually, users can describe what they want in plain language, and the system executes those steps across Photoshop, Premiere Pro, Lightroom, Illustrator and Express. The focus here is on reducing the need for step-by-step inputs. Users can give a broader instruction, and the assistant breaks it down into actions on its own.

Adobe is positioning this as a way to handle multi-step creative work in a more efficient way. For instance, tasks like editing visuals, adjusting audio or preparing content for different formats can be carried out within a single flow, without switching between tools manually. The assistant is also designed to retain context, meaning it can understand ongoing work and continue it across sessions or even across different applications.

To make this more usable in day-to-day workflows, Adobe has introduced pre-built “creative skills” for common tasks such as retouching images or generating content for multiple platforms. These workflows can be customised, or users can create their own depending on their needs. The assistant also works with existing assets like images, videos and brand elements, so edits are more context-aware rather than generic.

Importantly, Adobe has kept user control central to the experience. Even though the assistant can execute tasks end-to-end, users can review, refine or override changes at any stage, ensuring that the final output remains aligned with their intent.

Source link

Who's on the call: US bank CEO uses AI clone in quarterly earnings meet

by Hansraj Agrawal | Apr 28, 2026 | Technology

Tech companies are racing to adopt artificial intelligence (AI) in their day-to-day operations, but in an unusual move, a US-based bank has taken an early lead. Customers Bank CEO Sam Sidhu on Friday attended the first-quarter earnings call with analysts, and half an hour into the call, he revealed he hadn’t been speaking at all.

The CEO used his AI clone to deliver remarks at the quarterly conference. Calling it possibly the first-of-its-kind move for a public company’s earnings call, he said, “The prepared remarks you heard on my behalf today were delivered by my AI clone, not read by me,” CNBC reported.

His motive behind using his AI clone? To embrace a broader shift towards AI, the bank, headquartered in Pennsylvania’s Phoenixville, signed a multiyear deal with OpenAI to automate banking operations.

Increasing popularity of AI clones

This development comes as the adoption of AI clones is becoming more widespread, and companies are still navigating how to use them. Earlier this month, a Financial Times report stated that Meta is training its CEO, Mark Zuckerberg’s, AI avatar that could soon replace him in meetings. Zuckerberg’s AI avatar is reportedly being taught his mannerisms, tone, and public statements in a bid to make employees feel more connected to the founder. Zuckerberg is also reportedly creating an AI agent, which is separate from its avatar, to help him complete tasks.

But creating and replacing themselves with AI clones is not limited to CEOs, entrepreneurs, actors and influencers. More and more people are turning to new-age AI startups to help them create AI clones of their loved ones whom they’ve lost. In China, ‘ghost bots’ are gaining popularity, as tech companies are letting people purchase AI replicas of deceased loved ones, for a nominal fee.

Meanwhile, an Indian startup, Gnani.ai, is developing AI voice models that can clone people’s voices in 12 Indian languages.

But as the adoption of AI clones is becoming more widespread, so are concerns around fraud and misuse. Scammers are increasingly using AI-replicated voice technology to impersonate relatives or acquaintances and steal money. In a recent case, a Madhya Pradesh-based teacher lost ₹1 lakh after a fraudster allegedly posed as her cousin using AI to mimic his voice, tone and urgency.

While AI clones are being seen as important tools to streamline operations and help senior executives multitask, vulnerabilities and risks remain significant and largely unresolved.

Source link

OpenAI ends Microsoft exclusivity; frees path for Amazon, Google deals

by Hansraj Agrawal | Apr 28, 2026 | Technology

Microsoft and OpenAI renegotiated a pact that let Microsoft exclusively sell the ChatGPT creator’s artificial intelligence models, clearing the way for the startup to forge new deals with rivals to the software and enterprise giant, including Amazon.

The loosened ties between Microsoft and OpenAI, which have been expected for a while, are a sweeping change to one of the AI era’s most consequential alliances. It is arguably advantageous for both companies.

Microsoft shares initially fell 1.3 per cent on the news but closed largely unchanged. Alphabet closed up 1.81 per cent, while Amazon closed down 1.1 per cent.

Microsoft’s early investment, totalling $13 billion since 2019, helped pave the way for OpenAI’s ascent as an AI pioneer and powered growth in its Azure cloud computing business. Tensions between the two had been rising as OpenAI wanted the freedom to strike cloud deals with Microsoft’s rivals.

The renegotiated terms announced jointly will help OpenAI secure more computing power and build out an enterprise business that can better compete with Anthropic ahead of planned IPOs by both artificial intelligence companies. Microsoft will get greater certainty about its revenues from OpenAI under the deal, while OpenAI will gain newfound flexibility. Microsoft will remain OpenAI’s primary cloud partner with a license to the startup’s intellectual property through 2032.

Microsoft will also get a guaranteed 20 per cent cut of OpenAI’s revenue until 2030, though the total will now be subject to an undisclosed cap.

The fresh terms remove a rider that would have allowed OpenAI to stop paying Microsoft if it achieved so-called artificial general intelligence, the point at which AI matches or surpasses human ability.

In an internal memo reported by CNBC this month, OpenAI said the Microsoft partnership had been foundational but had limited the startup’s enterprise reach. The memo added that demand since OpenAI launched on Amazon’s cloud had been staggering.

“The new deal with Microsoft was essential for OpenAI to be successful in the enterprise market,” said Gil Luria, analyst at D.A. Davidson & Co. “AWS and Google Cloud enterprise customers have been limited in their ability to integrate OpenAI’s products because of the exclusive relationship and will now be more likely to consider OpenAI alongside Anthropic,” he added.

OpenAI’s promise to use at least $250 billion in Azure services by 2032 remains in place, with Microsoft having the right to make OpenAI products available first on Azure, unless Microsoft decides not to support them. Microsoft will also no longer pay OpenAI a share of Microsoft’s revenue for offering OpenAI models on Azure.

Signing deals with Microsoft rivals

The original deal gave Microsoft control over how OpenAI’s models were run on the cloud. Microsoft could offer the broadest access to OpenAI’s models, with cloud rivals such as Amazon able to offer only more limited and legally questionable ways of accessing the models.

The Financial Times reported last month that Microsoft was weighing legal action against Amazon and OpenAI over a $50 billion cloud deal that may breach its exclusive cloud tie-up.

The new deal does away with that.

In a post on LinkedIn, Amazon CEO Andy Jassy said OpenAI’s models would be available directly to developers on Amazon Web Services “in the coming weeks” and that the two firms would share more details at an event in San Francisco on Tuesday.

“With this, builders will have even more choice to pick the right model for the right job,” Jassy wrote. OpenAI has also struck deals including cloud and infrastructure agreements with Oracle and Alphabet’s Google, a chip partnership with Nvidia, and a manufacturing tie-up with Apple supplier Luxshare as it pushes into consumer devices.

Microsoft works to reduce OpenAI reliance

Microsoft appears to be allowing that deal to proceed and in exchange getting certainty around a lingering risk if OpenAI asserted it had achieved AGI.

Microsoft and OpenAI had restructured their tie-up in October, removing major constraints on the startup’s ability to raise capital and secure computing resources.

In recent months, Microsoft has been working to reduce its reliance on OpenAI by developing its own AI models and rolling out those developed by the likes of Anthropic in its products including the 365 Copilot for enterprises.

“From Microsoft’s perspective, it does not need to build out all the data center needs for OpenAI, freeing up capital for Copilot and other cloud capacity,” Barclays analysts said, calling the move a positive for both Microsoft and OpenAI.

Ending the exclusivity pact may help Microsoft fight antitrust scrutiny in the UK, the U.S. and Europe over whether its OpenAI tie-up gives it an unfair advantage in the cloud and enterprise AI markets.

Source link

« Older Entries

Next Entries »

Times Top News

Times Top News