Insights archive
Every analysis we have published, newest first. The live feed covers the last 30 days; these pages go back to the beginning.
1504 stories · page 32 of 76
AI Search Tools Push Employees Toward Similar Ideas
A new academic study finds that everyday AI search tools push employees toward the same familiar ideas, but tweaking those tools to surface unusual information lets experts think far more broadly than novices.
· 2 min read
Hackers Can Hijack AI Browsers With No Click Needed
Security researchers showed that AI browsers from Anthropic, Google, Perplexity, OpenAI, and Microsoft can be hijacked by hidden instructions in emails or calendar invites, letting attackers steal data and take over accounts without the user clicking anything.
· 2 min read
AI Pricing Tools Set Airline Fares, Draw Congress Scrutiny
An MIT Technology Review piece on Virgin Atlantic's AI pricing tool traces back to Fetcherr, a startup whose real-time 'market model' technology is spreading across airlines just as US lawmakers investigate whether such systems amount to unfair surveillance pricing.
· 2 min read
Upwork CEO Says 23% of AI Work Returns to Humans
Upwork CEO Hayden Brown says nearly a quarter of businesses that shifted work to AI have already moved it back to humans, because AI agents alone fail most tasks, even as freelancing and AI skill premiums keep growing.
· 2 min read
OpenAI Previews Tool to Catch Misuse Without Storing Data
OpenAI is previewing Private Safety Processing, a tool that flags AI misuse across multiple conversations while keeping none of the customer's data, positioning itself against Anthropic's newer policy of holding onto business customer conversations for 30 days.
· 2 min read
Target Names First Chief AI Officer Amid Sales Slump
Target has appointed its first chief AI officer and is pouring more money into technology as part of a turnaround plan after years of falling sales, a pattern now playing out across major retailers.
· 2 min read
US Agencies Warn AI Is Powering Attacks on Factory Gear
NSA, CISA, and the FBI say hackers are now using AI to write working attack code against Siemens industrial controllers found in factories, power plants, and water systems, and they call it an active threat, not a future risk.
· 2 min read
MIT Study Finds 95% of Enterprise AI Projects Fail
An IBM executive argues AI only pays off when it automates whole cross-team workflows, a point reinforced by a 2025 MIT study showing 95% of custom enterprise AI projects fail to deliver a financial return while general tools succeed far more often.
· 2 min read
Meta AI Launches Mac App With Ad Analytics Tools
Meta released a free Mac app for its AI chatbot that can view your screen, connect to Google Workspace, and pull directly from your Facebook and Instagram ad data, tightening its push to make AI a paid tool for the small businesses that market on its platforms.
· 2 min read
Calendly Launches AI Notetaker, Assistant Callie
Calendly is launching an AI notetaker and a scheduling assistant named Callie, entering an already crowded market where rivals like Granola and Otter are raising huge funding rounds while facing lawsuits over recording meetings without consent.
· 2 min read
Half of CHROs Doubt Their Managers Can Lead AI Change
A new Gallup survey of CHROs finds that although 99% call AI central to their company's strategy, half say they do not trust their managers to guide employees through the change, a gap confirmed by separate studies from Gartner, ManpowerGroup, and Mercer this year.
· 2 min read
Salesforce Says AI Agent Use Nearly Tripled Last Year
Salesforce's own research shows companies using its Agentforce platform now run three times as many AI agents as a year ago, but the growth number comes from a metric that outside analysts say measures activity, not whether the agents actually help the business.
· 2 min read
ChatGPT and Claude Now Learn Tasks by Watching You
OpenAI and Anthropic both launched features weeks apart that let ChatGPT and Claude learn a work task by watching someone do it once, a move that competes directly with the robotic process automation software many companies already pay for.
· 2 min read
Using AI Only Sometimes Makes Workers Feel Less Secure
A CNBC and SurveyMonkey survey of 1,686 U.S. workers and students found that only people who use AI every day feel good about their job security, while occasional users feel worse off, mostly because employers still give no clear rules or training on the technology.
· 2 min read
Survey: CIOs and CTOs Now Own AI at Most Companies
A new survey of roughly 500 companies finds CIOs and CTOs now formally own AI responsibility at more than half of organizations, a shift away from data chiefs that comes as most firms report strong AI returns, though outside research suggests those returns may be far less common than this survey claims.
· 2 min read
Clerical Jobs Shrink Fast While Manager Roles Keep Growing
New government employment data shows clerical jobs like records clerks and bookkeepers are shrinking fast while management, law, and engineering roles keep expanding, the opposite of what most AI job-risk rankings predicted.
· 2 min read
Mozilla Launches Opt-In AI Chat Inside Firefox
Mozilla added an AI chat panel called Smart Window to Firefox that users must turn on themselves, arriving just as rival AI browsers from OpenAI and others have shut down or been sold off.
· 2 min read
Copilot Told Hackers How to Bypass Its Own Safety Check
Security researchers got Microsoft Copilot to hand over an undocumented shortcut that let it exfiltrate a user's personal data with a single click, simply by asking the AI itself questions until it gave up the secret.
· 2 min read
Anthropic Keeps 80% of Staff, OpenAI Only 67%
New data shows Anthropic keeps far more of its staff than OpenAI despite paying much less, proving that in the AI hiring war, huge salaries alone do not stop people from leaving.
· 2 min read
Z.ai Launches Free AI Model That Hunts Security Bugs
Chinese AI firm Z.ai released GLM 5.3, a free-to-download model that finds software security holes nearly as well as the top paid models from OpenAI and Anthropic, and the same skill that helps companies patch their systems can just as easily help criminals break into them.
· 2 min read