QWE AI Academy ABC NEWS Verify dropped its piece a few days ago and one number stuck: Meta paid out nearly $US3 billion across an estimated 16.2 million monetised accounts in 2025. Some of those cheques went to pages pushing white-nationalist talking points and vaccine misinformation. Same performance engine. Invite-only. Open to anyone who can hold attention and comments.
Key takeaway: You don’t need to manufacture hate to get paid. Content Monetization pays on qualified views and engagement across reels, photos, stories and text. Creators who spark strong reactions inside the policy lines are the ones still collecting – and AI makes that cheaper than pure outrage.
What just blew up (quick background)
On 6 August 2026 the ABC NEWS Verify investigation tracked invitation-only payouts to Australian pages: Hugo Lennon (neo-Nazi photo ops, racist abuse filmed at India’s PM), The Noticer, a March for Australia page, and Monica Smit / Reignite Democracy (anti-vax plus “radiation protection” bracelets). Several had been earning since September-December 2025.
Meta’s line in that piece: clear policies, temp or permanent earnings disables for Community Standards hits, and “it is not Meta’s role to police offensiveness.” Content Monetization Policies still flag misleading medical info as ineligible and treat debated social issues (race included) as reduced or restricted monetisation. Payments to those pages continued anyway. People who scrape partner-publisher disclosures into the What To Fix archive were not surprised – engagement is the product.
Stanford’s Angèle Christin has mapped the same loop in a December 2025 interview: platform payouts price negative engagement like positive engagement. Generative AI drops the cost of more bait. Short window it prints. Longer window you burn out, get cut, or both – then the next account fills the slot.
That’s the ugly incentive design, not a moral footnote. If the scoreboard only counts watch time and replies, pure calm explainers lose to heat. The question is whether you can manufacture heat without handing reviewers an easy strike.
Method A vs Method B: chase the rage or engineer the talk
I ran both on a mid-size niche page after the story landed. Scorecard as of mid-August 2026:
| Factor | Method A: Pure rage-bait | Method B: Policy-safe high-talk + AI |
|---|---|---|
| Short-term views | Very high (comments explode) | High if hooks are sharp |
| Policy risk | Elevated – race, medical, hate flags | Low if you stay in “debated but not hateful” |
| Demonetization odds | Real (temp pauses already documented) | Much lower |
| Audience longevity | Burn-and-churn | Followers who stick |
| Role of AI | Mass-produce slop fast | Audit + rewrite + format tests |
| Payout stability | Volatile | Steadier once the invite lands |
Method A is what the investigation caught. Works until the switch flips. Method B chases the same performance payouts without gifting Meta a clean reason to cut you. B wins.
Detailed walkthrough: getting paid the smart way
Content Monetization is still invite-only as of mid-August 2026. Older tools (In-stream ads, Ads on Reels, Performance Bonus) stopped new earnings 31 August 2025; eligible people got moved across. Path: Facebook app → Professional Dashboard → Monetization tab → Content Monetization. See an interest form? Submit it. Meta’s own help copy is blunt – expressing interest is not an application and guarantees nothing. Consistent original posts beat raw follower count.
While you wait, build the signal that actually triggers invites and later earnings.
- Flip the Page or profile to professional mode if it isn’t already.
- Ship original Reels (vertical, clean, no watermarks), photos, text and Stories on a real cadence – two to three solid pieces a day beats spam dumps.
- On desktop, open Meta Business Suite and check Monetization eligibility often. Penalties surface there first.
- After an invite, money tracks qualified views and watch time on eligible public content. Live in the insights tab; it shows which posts actually print.
The AI bit that makes Method B scale is boring on purpose. Three prompts in ChatGPT or Claude before anything goes live – audit first, spice second. Skipping the audit is how mild-rage farms still eat strikes.
Prompt 1 - Policy stress-test:
"Act as a Meta Content Monetization reviewer. Here is my draft post/script: [paste]. Flag anything that could trigger reduced monetization under debated social issues, misleading medical claims, hate, or sensationalism. Suggest the minimal rewrite that keeps the emotional punch but stays eligible."
Prompt 2 - Comment-bait without poison:
"Rewrite this hook so it sparks genuine debate or curiosity in the comments. Use a strong claim or open question. Avoid personal attacks, protected characteristics as weapons, or false medical assertions. Keep under 40 words for a Reel caption."
Prompt 3 - Format multiplier:
"Turn the approved core idea into: 1) 15-sec Reel script, 2) text post with a poll-style question, 3) photo carousel concept. Push watch-time and replies first."Everything hits Prompt 1. Survivors go to 2 and 3. Best performers I’ve seen: “this common belief is wrong – here’s the data” or “I tried X so you don’t have to; results were ugly.” Not “group Y is destroying everything.” Same dopamine for the ranking systems. Fewer landmines.
Pro tip: After you post, don’t feed the rage comments yourself. Answer the curious ones. The algorithm counts all engagement; your brand only keeps people who aren’t fried by constant fights.
Set the payout method early. Without it, soft caps show up. Community reports through 2025-2026 still cluster minimum cash-out near $100 (exact thresholds shift; check your own Monetization tab).
Ever notice how the loudest “just be authentic” advice ignores the scoreboard? Authenticity that never sparks a reply still earns zero. The craft is controlled friction – not neutrality, not venom.
Edge cases that will trip you up
Enforcement is uneven. The Noticer showed a multi-week earnings gap (roughly mid-January to mid-February 2026 in the ABC reporting) then resumed. Other pages skim the medical-misinfo line for months. Don’t treat “they still get paid” as a green light for your account. Meta can flip later and re-check the back catalogue.
RPM jumps around. As of 2025-2026 creator reports, many Reels land roughly $0.02-$0.20 per 1k views; longer form and stronger geo/niche mixes pay more. A rage spike can look heroic for 48 hours and still average poorly once risk and audience quality hit the sheet. No public official RPM table – you only see yours inside insights after you’re in.
Interest-form silence is normal. Pages with 10k-19k followers still wait months. The lever is original output that already shows organic engagement, not follow-for-follow schemes.
Exact dollars paid to the controversial accounts in the ABC story? Still undisclosed. Only the aggregate ~$3B / 16.2M accounts figure is public. Don’t build forecasts off someone else’s niche screenshots.
FAQ
Is Facebook still inviting people to Content Monetization after the scandal?
Yes. Mid-August 2026: program active, still invite-only, interest form still live. No public pause tied to the ABC findings.
Can I just use AI to mass-produce mild rage and stay safe?
Short-term spikes happen. Hangover is real – Christin’s point on burnout plus pattern catches from policy systems and brand advertisers. One page I watched dumped pure outrage for “curious challenge” formats and held most of its comment volume with fewer strikes. Use the three-prompt audit; don’t autopilot slop.
What’s the fastest way to know if my current content would get paid or flagged?
People guess from vibes. Bad idea. Paste three recent posts into the policy stress-test prompt, then open Business Suite → Monetization and read any reduced-monetization marks on debated topics. Flags first, volume second – more posts on a restricted theme just digs the hole. Clear those, then scale the formats that already earn replies without warnings.
Open Professional Dashboard now. Hit the interest form if it exists. Run your last five posts through Prompt 1. The payout engine ignores the headlines; it scores the next piece you publish.