Copilot Cowork Takes the 9-to-5 Shift

Clock in, Tech Insiders.

Copilot just hired itself as your cubicle mate, Grammarly's channeling ghostwriters, and layoffs are turning water-cooler chat into exit-interview practice.

Punch the virtual timecard; today's shifts start below.

Here's what you need to know today:

Copilot Cowork Gives Microsoft 365 an Autonomous Assistant

Think of it as Claude in a power suit.

Microsoft unveiled Copilot Cowork, a capability that turns simple prompts into multistep plans executed across Outlook, Teams, Excel, and the rest of Microsoft 365.

Powered by Anthropic's Claude models and Microsoft's Work IQ intelligence layer, Cowork first drafts a step-by-step plan, then checks in for approval before rescheduling meetings, drafting decks, or updating documents. So you stay in control while the grunt work disappears.

Early demos show Cowork clearing jam-packed calendars, building complete meeting packets, compiling research, and even sketching full product-launch workflows. By tapping every email, chat, and file you've already stashed in the Microsoft Graph, the agent can act with surprising context; no copy-paste gymnastics required.

The tool was announced yesterday and is currently in research preview for a small group of enterprise customers, with broader access coming later this month through Microsoft's Frontier early-adopter program. Enterprise guardrails, including identity, permissions, and audit logs, apply automatically, aiming to soothe IT nerves about a bot that can move company data at will.

To get it, companies will need a $30-per-user Copilot license or Microsoft's newly announced $99-per-user E7 subscription tier launching in May.

Why it matters: For knowledge workers drowning in meetings and follow-up docs, Copilot Cowork could be the first mainstream "do-the-work-for-me" agent. If it delivers on Microsoft's promises, your next time-saver might not be a plug-in or macro—it'll be an AI colleague who never drinks the last of the office coffee.

Share this newsletter!

Pulse Check

Would you let Copilot Cowork rearrange your week?

Results from Yesterday's Pulse Check

Would you wear Samsung's screen-free AI smart glasses?

Insider Intel

Grammarly's 'Expert Review' Sparks Consent Uproar

Turns out even the dead can't escape revision notes.

Grammarly's latest AI agent, Expert Review, promises "feedback inspired by real experts," but it does so by labeling its suggestions with the names of authors, journalists, and academics—some of whom never agreed to help and, in at least one case, recently passed away.

The tool scans a user's draft, matches it to harvested public writings, and drops in edits as if they came straight from, say, Neil deGrasse Tyson or historian David Abulafia.

Early testers at The Verge found comments styled as personal notes from their own newsroom staffers, complete with profile-style blurbs, even though none had granted permission. Clicking the cited "sources" sometimes led to spammy mirror sites or unrelated pages, further muddying attribution.

Grammarly insists the agent merely draws on "publicly available" work and that any likeness is meant as inspiration, not endorsement.

Scholars aren't buying it. Academics on Bluesky dubbed the practice "digital necromancy," highlighting how the late Professor Abulafia appeared as an active reviewer little more than six weeks after his death.

Critics warn the feature commodifies reputations without consent, creating "little LLMs" of real people, and could trigger copyright or right-of-publicity fights.

For now, Expert Review remains opt-in, but if your draft suddenly channels Stephen King, don't be surprised when the living (or deceased) come knocking.

At least Strunk and Sagan aren't around to leave angry margin notes... yet.

Flash Bytes

Build a Resilient Epic Future

Healthcare leaders are reassessing Epic infrastructure as cloud capabilities mature. The challenge is balancing modernization with resilience and compliance.

Join TechnologyAdvice and UST on March 25, 2026, at 1:00 p.m. ET for Epic Infrastructure Modernization: A Strategic Roadmap to the Cloud. This session outlines a practical road map for isolated recovery environments, disaster recovery, and enterprise-scale Epic on cloud.

Evaluate your infrastructure strategy before your next major investment decision.

Security Alerts

Ericsson Breach Exposes IDs, Bank Data, and Health Info

Ericsson's US arm is warning staff and customers after a service provider hack exposed personal files, including SSNs, driver's licenses, medical information, and banking data between April 17 and 22, 2025.

Filings were made to the attorneys general of California and Texas, but only the number of affected Texans (4,377) is known at this point; Ericsson hasn't shared the total count or named the vendor, but says no misuse has surfaced yet.

Next steps: Enroll in Ericsson's free IDX credit monitoring before the June 9 deadline, freeze your credit, change shared passwords, and scrutinize bank statements for fraud.

OpenAI Codex Security Finds 11,353 Bugs in a Month

OpenAI pushed its Codex Security agent into research preview. The scanner tore through 1.2 million commits across external repos, flagging 10,561 high-impact and 792 critical vulnerabilities within 30 days.

Teams plug Codex, which runs on GPT-5.4, into GitHub, where it explains flaws, validates them in an isolated sandbox, and auto-generates pull requests. With findings staying secure in containers, OpenAI also launched Codex for Open Source (OSS) to give maintainers free access.

Next actions: Plug Codex into your CI/CD pipelines and scan historical branches to catch lingering flaws. Just keep a human reviewing those auto-generated patches, unless you want Skynet committing to prod.

Industry Shakeups

Tech Jobs Slide as AI Meets a Cooling Labor Market

February's jobs report surprised economists with a loss of 92,000 positions, including 11,000 shaved from the information sector.

Analysts say tech's three-year employment slide has surpassed the Great Recession's pace, though it remains shy of the early-2000s dot-com nadir.

Meanwhile, National Economic Council Director Kevin Hassett downplayed the carnage, telling CNBC the economy "is moving the fastest we've ever seen." Fast, sure... straight off a cliff.

Why the pain? Macro weakness plays a part, but AI is the convenient (and highly lucrative) scapegoat. Block just axed a staggering 40% of its workforce, citing its new AI agent, while Oracle is slashing thousands of roles to fund a $50 billion AI data-center spree. Morgan Stanley also blamed generative tools for recent cuts.

Unfortunately, displaced developers are finding out they can't simply cross the street to an LLM startup; AI companies are adding far fewer jobs than Big Tech once did.

What's next? Expect a slow "dribble of bad news."

Anthropic's new "observed exposure" index puts computer programmers in the hottest seat, with 75% of their tasks deemed AI-ready. Juniors are already being frozen out; entry-level tech hiring for 22- to 25-year-olds has dipped 14% since 2022.

Workers combining domain expertise with AI oversight will survive, while companies should budget for retraining over severance to preserve institutional knowledge.

Update your résumé before the bots label it "obsolete," then have them proofread it for you. At least until they automate that, too.

Meet Our Author

Justin Meyers

Writer at TechnologyAdvice

Justin Meyers is an investigative writer and editor who draws on over a decade of meticulous hands-on research to deliver the full, trustworthy story behind consumer and enterprise tech, including cybersecurity.

Curious about where AI is really headed?

The Neuron cuts through the noise to bring you smart, hype-free takes on the latest AI trends, tools, and breakthroughs. Join 500,000+ professionals from top companies like Microsoft, Apple, Salesforce and more.

Advertise in Daily Tech Insider!

Daily Tech Insider is a TechnologyAdvice business.

© 2026 TechnologyAdvice, LLC. All rights reserved.

TechnologyAdvice, 3343 Perimeter Hill Dr., Suite 215, Nashville, TN 37211, USA.