A look at the proliferation of fake LinkedIn profiles; LinkedIn's transparency report says the company blocked 58M+ accounts in H2 2022, up from 22M in H1
https://www.ft.com/... Tweets: Glenn Gabe / @glenngabe : 58M... That's it? ;) -> A look at fake LinkedIn profiles; LinkedIn's transparency report says it blocked 58M+ accounts in H2 2022, up from 22M “While systems exist to spot profile cloning & spam accounts, A.I. has made identification more difficult.” https://www.ft.com/... [image] Ian Fraser / @ian_fraser : The trouble with fake profiles on LinkedIn. Personally I'd have thought William Grace-Hunter is an obvious fake as he has zero connections. https://twitter.com/... Robert Smith / @bondhack : William Grace-Hunter is a “senior investment consultant” at BlackRock who is passionate about the investment case for THG. The only catch: he doesn't actually exist @BryceElder & @FD take a look at the murky underbelly of LinkedIn https://www.ft.com/... [image]
Context & Ripple Effects
Fake LinkedIn profiles have been a documented problem for years: researchers flagged over a thousand accounts with computer-generated pictures used for lead generation and product promotion back in March 2022, and by October the pattern had escalated to full [[a:983544|fake executive personas pairing AI-generated headshots with text scraped from real accounts]]. What changed with this report is scale: LinkedIn's own transparency numbers show enforcement climbing from 21.6M fake accounts blocked in H1 2019 to 22M in just H1 2022, then jumping to 58M+ in H2 2022.
The named example makes the threat concrete — a 'senior investment consultant at BlackRock' named William Grace-Hunter who does not exist. The report's key admission is that the systems built to spot cloning and spam are losing ground because AI has made identification harder.
First-order effects
- LinkedIn's detection systems now face a volume nearly triple its early-2022 rate in a single half-year, meaning more of its moderation budget is consumed at registration rather than post-hoc cleanup.
- Recruiters, HR departments, and invite-only groups — already dealing with fake executive profiles per prior coverage — must treat connection requests and consultant claims like BlackRock affiliations as unverified until proven real.
Second-order effects
- Lead-generation operations that abused networks of AI-photoed profiles get a cheaper toolset as generation improves, forcing sales teams to re-weight outreach away from raw LinkedIn volume toward verified channels.
- Employers screening candidates absorb rising verification costs, since a plausible-looking profile with zero connections (as commenters noted about the Grace-Hunter account) becomes an unreliable signal of legitimacy.
Third-order effects
- If the pattern holds — including later analysis finding most long English-language posts on the platform likely AI-written while LinkedIn says it doesn't track AI usage — the professional network's value shifts from content and connections to verifiable identity, pushing platforms toward provenance checks rather than spam filters.
- Trust in open professional networking erodes structurally, favoring smaller, invite-vetted communities and third-party identity verification as complements to platform-native moderation.
The trend: Generative AI is turning platform integrity from a spam-volume problem into an identity-problem, with LinkedIn's blocked-account counts tripling as synthetic profiles outrun content-based detection.