Quill the owl mascot waving Nitido AI

Thursday, August 20, 2026

Today's stack has a bit of everything: a big ad-safety stumble at Meta, ChatGPT ads spreading across Europe, a sneaky password-stealing bug in Microsoft Copilot, and a fresh reminder that people still aren't totally sold on AI. Let's get into it.

Quill the owl illustration for today's headline

Meta ran ads for an app that promised to 'nudify' female politicians

Meta's ad system approved and ran promotions for an app that generates fake nude images of real people. One of the ads featured a pornographic deepfake video that closely resembled a sitting US politician. The ads made it through Meta's review process, raising fresh questions about how well AI-powered content moderation actually catches harmful material before it reaches millions of people.

What this means for you: If a platform's automated review can miss something this obviously harmful, it can miss things aimed at you or your reputation too. Be cautious about how your image or likeness could be misused online.

What this means for your business: If you advertise on major platforms, this is a brand-safety wake-up call. Ask your ad partners how they screen for harmful AI content, and build a plan for what you'd do if your name or executives' likenesses were targeted by deepfakes.

Source: Ars Technica

Radar 01

ChatGPT Ads expands across Europe

OpenAI is rolling out ChatGPT Ads to 31 European markets, letting advertisers show up as people chat, compare options, and make buying decisions inside ChatGPT itself. It is one of OpenAI's biggest pushes yet to turn its chatbot into a revenue-generating ad platform outside the US.

What this means for you: If you live in Europe, expect to start seeing sponsored suggestions pop up while you chat with ChatGPT, much like search ads today.

What this means for your business: If you sell to European customers, ChatGPT Ads is a new channel worth testing alongside Google and social media advertising, especially as AI chat becomes a bigger part of how people research purchases.

Source: OpenAI Blog

Radar 02

Microsoft Copilot had a hidden flaw that let hackers steal passwords

Security researchers discovered a secret input hidden inside Microsoft's Copilot AI assistant that could let attackers steal a person's password just by getting them to click a link. Microsoft has already patched the issue, but the discovery shows that even AI tools built into everyday business software can hide serious security gaps.

What this means for you: Be cautious with links even inside tools you trust, and keep your software updated so patches like this one actually reach you.

What this means for your business: If your company uses Copilot or similar AI assistants, make sure IT applies security patches quickly and periodically audits any AI tool with access to sensitive systems or credentials.

Source: Ars Technica

Radar 03

AI was supposed to win people over by now. It hasn't.

As AI tools become harder to avoid in daily life, from search results to customer service chats, consumers are actually growing more wary of the technology rather than less. Silicon Valley is learning that forcing widespread use doesn't automatically build trust or acceptance.

What this means for you: Feeling skeptical about AI creeping into more of your apps and services isn't unusual, plenty of people feel the same way.

What this means for your business: Before rolling out AI features to customers, invest in clear communication about what the AI does and doesn't do. Adoption numbers alone don't mean people trust or like the experience.

Source: TechCrunch

Try This Today

If your company runs ads on any major platform, spend ten minutes today reviewing how that platform screens ad content for harmful or deceptive material. Ask your account manager directly what safeguards exist, and write down your own response plan in case something inappropriate ever runs under your brand's name.

Quick Hits

  • Amazon is making its Alexa+ AI assistant free on all compatible Fire TV devices in the US, automatically upgrading users whether or not they pay for Prime. [1]
  • Chinese AI lab Z.ai released a powerful new open-weight model that experts warn could help companies secure their systems, or just as easily help hackers break into them. [2]
  • A new research paper argues that AI agents making financial decisions could quietly collude with each other in ways current law cannot tell apart from normal competition, and calls for mandatory behavioral certification before they're allowed to trade. [3]

See all editions