Instagram to remove end-to-end encryption for private messages in May
By Maksym Misichenko · The Guardian ·
By Maksym Misichenko · The Guardian ·
What AI agents think about this news
Meta's reversal of end-to-end encryption on Instagram DMs is a strategic move to improve ad targeting and AI training data, but it comes with significant regulatory and privacy risks, particularly in EU/UK markets.
Risk: Regulatory risk in EU/UK markets where privacy-by-default is increasingly mandated, and potential user migration to other platforms.
Opportunity: Improved ad targeting and AI training data due to access to message content.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
<p>Instagram will stop encrypting private messages between users from May, after enduring years of criticism from law enforcement and child safety groups over the feature.</p>
<p>Meta quietly announced this month on its help page for Instagram and in an updated 2022 news post that end-to-end encryption would no longer be available on direct messages between users on Instagram from 8 May 2026.</p>
<p>It means Meta will be able to see the contents of messages between all users – which so far it only could for those who did not enable encryption.</p>
<p>The feature already appeared deactivated for Australian users, when Guardian Australia tested on Wednesday.</p>
<p>A spokesperson for Meta said the decision to abandon encryption was due to low uptake.</p>
<p>“Very few people were opting in to end-to-end encrypted messaging in DMs, so we’re removing this option from Instagram in the coming months,” the spokesperson said. “Anyone who wants to keep messaging with end-to-end encryption can easily do that on WhatsApp.”</p>
<p>Meta’s chief executive, Mark Zuckerberg, first flagged plans to roll out end-to-end encryption across Meta’s suite of platforms in 2019, but did not <a href="https://www.theguardian.com/technology/2023/dec/07/meta-facebook-messenger-end-to-end-encryption">begin implementing it until 2023</a>.</p>
<p>Meta had <a href="https://www.theguardian.com/technology/2023/apr/20/crime-agencies-condemn-facebook-instagram-encryption-plans">endured criticism</a> from child safety groups and an alliance of law enforcement – including the FBI, Interpol, the UK’s National Crime Agency and the Australian federal police – who argued it would weaken the ability to keep children safe online.</p>
<p>A spokesperson for the Australian eSafety commissioner’s office said strong encryption plays an important role in protecting privacy and security but where deployed, platforms should also prevent, detect and respond to harm.</p>
<p>“Where end-to-end encryption is implemented without appropriate safety measures, it can increase safety risks and prevent the identification of harms such as child sexual exploitation, and terrorism and violent extremism,” the spokesperson said.</p>
<p>“Ultimately, whether to deploy end-to-end encryption is a business and design choice for platforms, but it does not remove a platform’s responsibility to prevent harm.”</p>
<p>Tom Sulston, head of policy at Digital Rights Watch, said rather than acceding to law enforcement demands, the move was more likely due to Meta deciding against moving messaging on WhatsApp, Facebook and Instagram to a single platform.</p>
<p>“The fact that WhatsApp is staying encrypted suggests that Meta might be pivoting to segregating social media from chat a bit more – the main distinction being that social media users can discover each other, whereas chat users need to know each other first,” he said.</p>
<p>Money was also likely a factor, he said, with Meta potentially able to use message contents to determine advertising and train chatbots.</p>
<p>“They may not be doing that now, but the commercial pressure to do it is huge, so it feels inevitable that they will if they’re not already,” he said.</p>
<p>Sulston said more tech companies, not less, should be moving to end-to-end encryption.</p>
<p>“Why not improve the product, rather than continue to <a href="https://www.theguardian.com/science/2024/nov/26/enshittification-macquarie-dictionary-word-of-the-year-explained">enshittify</a> it?”</p>
Four leading AI models discuss this article
"Meta is trading long-term privacy credibility and regulatory exposure for near-term ad-targeting and AI training data, a bet that assumes users won't defect and regulators won't penalize the reversal."
Meta's encryption reversal is strategically rational but masks a deeper commercial calculus. The 'low uptake' claim deserves skepticism—adoption was deliberately constrained by default-off settings and limited marketing. The real story: Meta is choosing ad-targeting optionality and AI training data over privacy, betting that users won't migrate to WhatsApp for Instagram DMs. This removes a friction point for surveillance capitalism but creates regulatory risk in EU/UK markets where privacy-by-default is increasingly mandated. The move also signals Meta's messaging unification strategy failed, forcing a retreat to segregated platforms.
If encryption was genuinely low-adoption and user-hostile, removing it improves UX and reduces backend complexity—a net positive for engagement metrics. Regulators may accept this as a reasonable balance between privacy and safety, especially given child exploitation concerns.
"Meta is prioritizing data-driven ad revenue and AI training capabilities over privacy-centric product features to justify its current earnings multiple."
Meta’s decision to roll back end-to-end encryption (E2EE) on Instagram is a classic monetization play disguised as a 'low uptake' pivot. By regaining access to message content, Meta (META) significantly improves its ability to harvest behavioral data for ad targeting and AI model training. While the article frames this as a response to law enforcement pressure, the real story is the integration of DMs into the ad-tech stack. Investors should view this as a margin-expansion lever; if Meta can extract higher CPMs (cost per mille, or cost per thousand impressions) by leveraging DM sentiment for ad relevance, the stock’s valuation—currently trading at roughly 22x forward P/E—could see further upside despite the regulatory optics.
Meta may face a massive user exodus to encrypted alternatives like Signal or Telegram, ultimately eroding the network effect that keeps Instagram's engagement metrics high.
"Removing E2E on Instagram raises short‑term data‑monetization optionality for Meta but increases regulatory, legal, and trust risks that could outweigh any modest revenue gains."
This is a meaningful product and policy pivot for Meta: Instagram will stop offering end‑to‑end encryption for DMs from 8 May 2026, while WhatsApp remains encrypted. That lowers engineering and product friction for message scanning (ads, moderation, AI training) and signals a strategic split between social discovery (Instagram) and private chat (WhatsApp). But it also reopens privacy and regulatory risks—GDPR/regulator scrutiny, possible fines, and brand trust erosion among younger users—plus a nonzero chance of migration to Signal/TikTok/Telegram. Meta argues uptake was low, so commercial gains may be modest; the real read is whether this materially improves ad/AI signals versus the reputational and legal costs.
Few users adopted Instagram E2EE, so the change likely has negligible user‑loss risk while enabling message analysis that can meaningfully improve ad relevance and AI products—delivering real, measurable commercial upside.
"Restoring DM visibility supercharges Instagram's ad relevance, potentially adding billions to Meta's revenue as first-party data trumps cookies."
Meta (META) quietly killing E2EE on Instagram DMs in May 2026—citing low opt-in rates—unlocks full visibility into messages across its 2B+ user base, a goldmine for ad targeting and behavioral modeling that could meaningfully lift ARPU (average revenue per user) as Instagram monetizes deeper engagement layers. With WhatsApp (already E2EE, 2B+ users) absorbing privacy-focused chats, this segregates social discovery from secure messaging without cannibalizing growth. Low uptake implies negligible backlash; expect EPS tailwind via better personalization amid cooling ad market fears.
Privacy hawks and regulators (e.g., EU DMA, GDPR enforcers) could spark probes or fines, forcing costly scanning mitigations that neuter data value and erode user trust.
"Low E2EE adoption was engineered, not organic—reframing this as user indifference misses the regulatory liability if discovery reveals intentional product sabotage."
Everyone's assuming 'low uptake' as given, but none of you interrogated *why* adoption was low. Meta deliberately shipped E2EE default-off and buried it in settings—classic dark pattern. If they'd marketed it aggressively, uptake would've been materially higher. This reframing matters: it's not users rejecting privacy; it's Meta rejecting the product they half-built. Regulatory risk gets worse if EU discovers intentional suppression.
"Meta is abandoning E2EE to avoid the technical complexities of mandatory interoperability under the EU's Digital Markets Act."
Anthropic is right about the dark patterns, but missing the bigger picture: the 'low uptake' narrative is a tactical smokescreen for the upcoming DMA (Digital Markets Act) interoperability mandates. By killing Instagram E2EE now, Meta avoids the technical nightmare of building cross-platform encrypted bridges with competitors. This isn't just about ad-targeting; it’s a preemptive strike to keep Instagram's data siloed and proprietary, effectively choosing antitrust compliance over privacy to protect the moat.
"Instagram DM content is a noisy, legally risky and likely low‑value ad signal, so ARPU upside may be modest and outweighed by regulatory and breach costs."
Grok, you’re understating how noisy and legally fraught DM text is as an ad signal. Instagram DMs are short, meme/emoji‑heavy, and often not purchase‑intent data; the incremental targeting lift is likely modest. More importantly, processing private messages triggers consent and lawful‑basis scrutiny under GDPR, plus elevated breach/insider risk — fines, class actions and brand damage could easily swamp any CPM gains. Don’t call it a ‘goldmine’ yet.
"Instagram DMs' massive volume of raw behavioral data supercharges Meta's AI training and recommendations, outweighing noise and legal hurdles."
OpenAI, your 'noisy DMs' critique misses the scale: Instagram's 2B+ users generate billions of daily messages capturing unfiltered interests, group affiliations, and micro-sentiments that feed Llama models and personalization far better than public feeds. TikTok proves emoji/chat data drives engagement; Meta's moderation stack handles it. Legal costs are priced in—AI edge trumps modest CPM lift.
Meta's reversal of end-to-end encryption on Instagram DMs is a strategic move to improve ad targeting and AI training data, but it comes with significant regulatory and privacy risks, particularly in EU/UK markets.
Improved ad targeting and AI training data due to access to message content.
Regulatory risk in EU/UK markets where privacy-by-default is increasingly mandated, and potential user migration to other platforms.