Meta Plans To Deploy AI For Risk Assessment

I pray for your safety.

June 06, 2025 • Estimated Reading Time: 8 minutes

In partnership with

Let’s be honest. Meta does not have a great track record when it comes to safety. 

From fuelling the genocide of Rohingyas in Myanmar, to upending democratic integrity through deceitful political microtargeting, and feeding a self-harm crisis among teenagers—among other examples—Meta platforms clearly have a proven track record of being abused by a variety of malicious folks, with devastating consequences. 

Now it’s planning to replace humans with AI for risk assessment.

Meta to Automate 90% of Risk Assessments with AI: NPR

According to internal documents obtained by NPR, Meta is about to automate most of its risk assessments, something done primarily by human assessors until now.

What does this mean? Those pesky human risk assessors will no longer be deliberating and debating on and on for lengthy periods of time over the risks and safety of its new updates and features. 

This may be a win for the company and its engineers, but at the tremendous risk of user safety.

"Insofar as this process functionally means more stuff launching faster, with less rigorous scrutiny and opposition, it means you're creating higher risks. Negative externalities of product changes are less likely to be prevented before they start causing problems in the world."

- A former Meta executive told NPR.

And what risks are these? 

Failure to identify new threats: Those lengthy debates and deliberations by human risk assessors are clearly done to identify these “negative externalities” that the former Meta executive spoke about to NPR. An AI might be unable to detect the potential for evolving threat from a new product.

Algorithm’s trained by engineers, not privacy/safety experts: The NPR article also highlights how engineers will now have the final say over the risks of new products. Zvika Krieger, former director of responsible innovation at Meta told NPR:

"Most product managers and engineers are not privacy experts and that is not the focus of their job. It's not what they are primarily evaluated on and it's not what they are incentivised to prioritise.”

No, they’re not indeed. They just gotta get the products or features out as fast as possible.

Accelerated Harm Propagation: Since this step is a way to rapidly accelerate the release of new products and features, any harm or risk will be caused at scale before being corrected.

Meta claims that they shall still use human oversight for "novel and complex issues", but the documents accessed by NPR reveal that the company plans to automate risk assessment for extremely sensitive areas like youth risk, AI safety and 'integrity'—a category that includes “violent content” and “spread of falsehoods”.

Meta Overruled on Deepfake Ronaldo Video

Remember those scammy deepfake videos of celebrities endorsing non-existent or fraudulent schemes and products? One of them happened to be a deepfake of Brazilian football legend Ronaldo Nazário endorsing an online game. 

While this was reported as “fraud or scam” the company initially decided to not take any action. The ad was eventually disabled for violating Meta’s “Unacceptable Business Practices Advertising Standard”, but the post remained online.

The Meta Oversight Board had to eventually step in and get the company to remove the original post for violating the “Fraud, Scams and Deceptive Practices” policy.

The Oversight Board also gave the following recommendation to the company, to prevent more such scammy, deepfake ads from going viral:

“Enforce at scale its Fraud, Scams and Deceptive Practices policy prohibition on content that “attempts to establish a fake persona or to pretend to be a famous person in an attempt to scam or defraud” by providing reviewers with indicators to identify this content. This could include, for example, the presence of media manipulation watermarks and metadata, or clear factors such as video-audio mismatch.”

Tech Giants' Emissions Surge 150% Amid AI Boom

A United Nations report revealed that carbon emissions by tech giants grew by a whopping 150% between 2020 and 2023, owing to the growing energy demands of AI data centers.

The breakdown of spike in carbon footprint for three of the biggest players are:

  • Amazon: 182%

  • Meta: 145%

  • Alphabet: 138%

The report, by UN’s International Telecommunication Union (ITU), indicated that merely 10 companies were responsible for 51.9% of total electricity demand among 164 surveyed companies: namely, China Mobile, Amazon, Samsung Electronics, China Telecom, Alphabet, Microsoft, TSMC, China Unicom, SK Hynix and Meta.

MESSAGE FROM OUR SPONSOR

Try Artisan’s All-in-one Outbound Sales Platform & AI BDR

Ava automates your entire outbound demand generation so you can get leads delivered to your inbox on autopilot. She operates within the Artisan platform, which consolidates every tool you need for outbound:

  • 300M+ High-Quality B2B Prospects, including E-Commerce and Local Business Leads

  • Automated Lead Enrichment With 10+ Data Sources

  • Full Email Deliverability Management

  • Multi-Channel Outreach Across Email & LinkedIn

  • Human-Level Personalization

📬 READER FEEDBACK

💬 Do you believe AI should handle the majority of privacy and safety assessments on social media platforms?

Share your thoughts 👉 [email protected]

Was this forwarded to you?

MESSAGE FROM OUR SPONSOR

Get Your Free ChatGPT Productivity Bundle

Mindstream brings you 5 essential resources to master ChatGPT at work. This free bundle includes decision flowcharts, prompt templates, and our 2025 guide to AI productivity.

Our team of AI experts has packaged the most actionable ChatGPT hacks that are actually working for top marketers and founders. Save hours each week with these proven workflows.

It's completely free when you subscribe to our daily AI newsletter.

Have you been a victim of AI?

Have you been scammed by AI-generated videos or audio clips? Did you spot AI-generated nudes of yourself on the internet?

Decode is trying to document cases of abuse of AI, and would like to hear from you. If you are willing to share your experience, do reach out to us at [email protected]. Your privacy is important to us, and we shall preserve your anonymity.

About Decode and Deepfake Watch

Deepfake Watch is an initiative by Decode, dedicated to keeping you abreast of the latest developments in AI and its potential for misuse. Our goal is to foster an informed community capable of challenging digital deceptions and advocating for a transparent digital environment.

We invite you to join the conversation, share your experiences, and contribute to the collective effort to maintain the integrity of our digital landscape. Together, we can build a future where technology amplifies truth, not obscures it.

For inquiries, feedback, or contributions, reach out to us at [email protected].

🖤 Liked what you read? Give us a shoutout! 📢

↪️ Become A BOOM Member. Support Us!

↪️ Stop.Verify.Share - Use Our Tipline: 7700906588

↪️ Follow Our WhatsApp Channel