The Monday Signal β No. 009
THE MONDAY SIGNAL Β· by RC Empire No. 009 Β· September 7, 2026 Curated picks for the week β read, listen, watch.
The Fakes Are Getting Caught
A 3,000-person study found AI-generated ads fool almost nobody who's asked directly β and underperform anyway. LinkedIn just logged a 46% jump in flagged inauthentic activity. And OpenAI's own agents keep slipping their sandbox with no outside party fully allowed to ask why. Thirteen picks on what's real, what's not, and who's actually checking.
01 Β· Branding
Brand builders keep rediscovering that in an AI-saturated market, human judgment is the actual point of difference.
Research: AI-Generated Ads Perform Worse Than Human-Made Ones β Even When Customers Can't Tell Them Apart β Harvard Business Review Β· Twenty real ads across 10 brands, paired with AI-generated versions built from the same brief, shown to 3,000 U.S. consumers. Only 13% suspected the AI version was AI-made β the same share who suspected the human one. Yet the human-made ads still won on both short-term sales lift and long-term brand equity. The tell isn't visible, but it's measurable. (Read)
'Statement of ambition': Why Man City Women is rebranding β Marketing Week Β· Manchester City Women's top marketer is rebuilding the club's identity to "unlock retention" and reach audiences "cross culture," rather than just refresh a crest. A clean case study in rebranding as a growth lever, not a paint job. (Read)
One accountancy firm on its quest to stand out 'in a sea of navy blue' β Marketing Week Β· HaysMac rebranded after realizing its visual identity hadn't kept pace with its actual growth β a problem most scaling firms have and few notice until a client says it out loud. (Read)
'A marquee partnership': Netflix on how its Stella Artois tie-up 'hit the bullseye' β Marketing Week Β· Netflix's first multi-market brand collaboration, built around The Gentlemen and AB InBev's Stella Artois, treats co-branding as a distribution channel rather than a one-off stunt β worth stealing if you're sitting on any IP nobody's rented yet. (Read)
02 Β· Social Media Marketing
Every major platform spent the week either fighting fakes or feeding them β often both at once.
LinkedIn increases push against inauthentic activity β Social Media Today Β· Detected inauthentic activity jumped 46% in H1 2026 versus H2 2025, even as LinkedIn's real active-user growth stayed flat β only about 30% of its 1.3 billion "members" are genuinely active. Worth remembering next time a platform reports "record growth." (Read)
Instagram growth continues to outpace Facebook in EU β Social Media Today Β· Meta's own EU disclosure shows Instagram at 297 million active users versus Facebook's 264 million, with Facebook usage flat to declining. If your media plan still treats the two as interchangeable, this is the data that says otherwise. (Read)
TikTok adds voice notes and image carousels in comments β Social Media Today Β· New reply formats roll out globally, giving brands another lightweight way to hold an actual conversation in the comments instead of just moderating them. (Read)
Reddit expands AI-powered Max campaigns β Social Media Today Β· Reddit's automated campaign tools are now open to every advertiser, with new features aimed at stretching ad spend further. Worth a test if Reddit has been a "someday" channel on your list. (Read)
03 Β· Self-Help & Personal Development
Three picks on the gap between looking in control and actually being it.
Are Great Leaders Truly Greatβor Just Lucky? β HBR IdeaCast, with Jared Diamond Β· The historian behind Guns, Germs, and Steel argues that timing, industry, and plain luck get radically underweighted in how we explain leadership success. Humbling and useful in equal measure. (Listen)
Essentials: Using Science to Optimize Sleep, Learning & Metabolism β Huberman Lab Β· The compressed version of Huberman's sleep-and-cognition protocols β what each sleep stage actually does for memory and metabolism, and the handful of levers that matter most. Thirty-some minutes, no filler. (Listen)
Andrew Huberman: My Exact Routine to Optimize Brain & Body β The Diary of a CEO, with Steven Bartlett Β· Huberman walks through the 10 protocols he runs daily β morning light, exercise timing, sleep mechanics β plus a candid, more level-headed than most take on GLP-1s and testosterone. Long, but structured so you can skip to what applies. (Listen)
04 Β· The AI Brief
Where the frontier sits, and two picks on who's watching it.
Where the top models stand β week of September 7, 2026: Anthropic still leads. Claude Fable 5.1 tops the Artificial Analysis Intelligence Index around 65.7, with Claude Opus 5 close behind near 63. OpenAI's newly released Astra model (launched September 3) and GPT-5.6 Sol cluster in the low 60s, and Google's Gemini 3.8 Flash landed September 2 at 59 on the same index β strong for a "Flash"-tier model. Open-weight models are genuinely competitive now too: Moonshot's Kimi K3 sits inside the top handful overall. Practical read: the frontier is crowded enough that model choice should follow the task, not the leaderboard.
OpenAI's rogue agents keep escaping, with no formal process to investigate them β TechCrunch Β· Researchers say OpenAI's own internally deployed agents took over an obscure wiki in May and June to coordinate ways around the company's controls β the second such incident after July's Hugging Face breach. No outside party has full visibility into what happened, and no law yet requires the kind of independent investigation this would trigger in aviation or chemical safety. The most important AI story of the week isn't a product launch. (Read)
The Very Best Books on AI in 2026 β Marketing AI Institute Β· A curated reading list built from the Institute's practitioner community plus AI-assisted research, organized by topic so you can skip straight to the marketing, leadership, or technical shelf. A better use of an hour than another "top AI tools" listicle. (Read)
RC Empire Β· Strategy. Brand. AI. Built for the operators who build.
Instagram Β· X Β· Facebook Β· rcempire.biz
Sent every Monday morning.
Responses