If you block ads in Chrome, you have about a month. On 31 August 2026 Google clears every remaining Manifest V2 extension out of the Chrome Web Store, and uBlock Origin — 11,000,000 users, 4.7 stars from 35,500 ratings — is one of them.
Nobody has panicked because nothing looks wrong. We loaded the listing on 29 July 2026: live, install button working, no warning banner, last updated 8 July 2026. You will also read the opposite: a rival blocker’s blog says uBlock Origin is already gone from Chrome’s store. Untrue. The real position is odder — it is a Manifest V2 extension, Chrome has had Manifest V2 disabled since 2025, and the listing goes away at the end of August.
So the advice has changed — just not in the direction most articles say, because they get two things wrong: what Chrome actually lost, and how much better Manifest V3 has got since they were written.
1. What actually happened?
Under Manifest V2 a blocker used blocking webRequest: Chrome handed the extension every network request and waited while its JavaScript decided. Manifest V3 replaces that with declarativeNetRequest, where the extension registers rules in advance and the browsermatches them — the blocker stops being a gatekeeper and becomes a rule-writer. Google’s stated reason, on the Chromium blog in May 2024: the old model made extensions proxy all your traffic in order to filter it, at a cost to performance and privacy.
The dates matter, because this deadline moved repeatedly and most commentary quotes a superseded version. From Google’s own timeline: the store stopped taking new public MV2 submissions in January 2022. The date usually cited as the shutdown, 3 June 2024, was only warning banners on the Beta, Dev and Canary builds — nothing stopped working. Disabling on stable began 9 October 2024. On 31 March 2025 MV2 was off by default but could still be switched back on, and that hatch closed in Chrome 138, which reached stable on 24 June 2025 while Google’s timeline dates the change complete on 24 July 2025.
The tail end went unreported: Chrome 150 on 30 June 2026 and Chrome 151 on 28 July 2026 removed the last enterprise escape hatches, per a notice a Chrome engineer posted to the W3C WebExtensions group — MV2 is no longer allowed in any supported Chrome version, and the team will not keep the code behind a compile flag.
2. What did it genuinely cost?
The best account is uBlock Origin author Raymond Hill’s, in the FAQ for uBlock Origin Lite — his own Manifest V3 blocker. The headline consequence is not about ads: uBOL “will be less effective at dealing with websites using anti-content blocker”. What cannot survive as a pre-registered rule is anything needing a decision at request time: strict first- and third-party blocking, entity-wide filters like google.*, redirect rules, click-to-load, much regex filtering. Scriptlet injection is unreliable too, since an MV3 background page is a service worker the browser can suspend.
One you would notice as a user: generic cosmetic filtering — the part that hides the leftover box where an ad was — is off by defaultin uBOL, where it is on in full uBO. The rule budget is finite too, since Chrome’s API reference guarantees each extension only 30,000 static rules.
Hill’s verdict is blunt: uBOL is “too different from uBO to be an automatic replacement”; asked whether uBO would move its users across automatically, his answer is “No”; and he tells Chrome users to “find an alternative to uBO before Google Chrome disables it for good.”
Now the correction, because this is where nearly everything you will read is wrong. Chrome did not lose HTML filtering, response-body rewriting, CNAME uncloaking or IP-address filtering. Chrome never had them.They are Firefox-only, and Firefox-Manifest-V2-only, and listed as such on Hill’s own “uBlock Origin works best on Firefox” page. The gap between uBO on Firefox and any blocker on Chrome is real — it just predates Manifest V3 by years.
3. The part nobody reported: Manifest V3 got better
The platform described in the 2024 and 2025 commentary is not the one you would install onto today. Declarative rules gained response-header conditions in Chrome 128 and top-domain conditions in Chrome 145 — the second mattered enough that Hill revised his own FAQ, moving per-site switches and dynamic filtering from impossible to “theoretically partially possible”.
The bigger change is dated June 2026 in that same FAQ: uBlock Origin Lite now supports custom filters andsubscriptions to remote filter lists — the two capabilities every “MV3 cripples ad blocking” piece names as the fatal losses. One catch: it needs Chrome’s user-scripts API, which Google requires behind a user-facing toggle, so you must switch on “Allow User Scripts” yourself — it is not on by default.
AdGuard got there earlier — custom filters returned in v5.2.77 on 22 September 2025, and lists began updating independently of the extension in v5.4 on 7 May 2026. And uBOL shipped a release every two to five days through June and July 2026, so its filter lag is days, not weeks. Most articles a search hands you describe a platform 18 to 24 months out of date.
4. Does the difference show up in real browsing?
Someone measured it properly, once. Lukic and Papadopoulos at Goethe University Frankfurt ran four blockers as MV2/MV3 pairs — including uBlock Origin 1.59.0 against uBlock Origin Lite — across 924 websites, launching on 15 May 2025 and publishing in PoPETs 2026. On ads blocked: no significant difference(Δ = −0.04, p = 0.41). On trackers the MV3 versions blocked 1.8 more per site(p < 0.001). uBO Lite matched full uBO, and several MV3 blockers at once beat any single one.
Before concluding the controversy was noise, read the method — the authors are careful about it. They excluded sites that “deployed ad-blocking prevention or anti-automation mechanisms”, focused on display ads, used default settings, and note that automation does not replicate real interaction. No extension author has published a rebuttal; that limitation is in the paper’s own text.
Sound design for measuring ordinary display advertising — and it also draws the boundary of the finding, because it sets aside exactly the adversarial cases where the gap is expected to live. So: on ordinary ad-supported sites Manifest V3 is as good as Manifest V2, and the difference lives in the fights (anti-adblock walls, video ads), which is precisely where uBlock Origin earned its reputation.
5. So what should you install?
On the desktop, the honest answer is to change browser.Firefox plus uBlock Origin is the only mainstream combination where the blocker is unconstrained: Mozilla committed to keeping blocking WebRequest in May 2022, said in March 2024 that it has no plans to deprecate MV2 and would give at least 12 months’ notice if that changed, and in February 2025 — the most recent statement we found — restated support for both APIs, naming uBlock Origin and the Chrome users about to lose it. Switching browsers is our recommendation, not Hill’s; he leaves that choice to you.
Staying on Chrome: uBlock Origin Lite, with “Allow User Scripts” on. Same author, free, MV3-native, and it matched full uBO on ordinary display ads in that study. Its author still says it is not an automatic replacement — but it is the best thing available inside Chrome.
On Brave, use Shields and leave it there. Shields is immune to all of this, patched natively into Chromium rather than shipped as an extension. But do not count on Brave to keep full uBO: the Chromium 150 rebase forcibly disabled the four MV2 extensions Brave hosts itself, with no way to switch them back on (Brave issue #56655, 25 June 2026). Fixes landed in early July, but nobody we can point to has verified an install since, and Brave now maintains those patches alone.
On Safari, relax — nothing was taken.Safari’s content blockers have worked this way since 2015: rules compiled ahead of time, and per WebKit the extension never sees the URLs you browsed to — a stronger privacy guarantee than the one Google described. It supports CSS element hiding, so the gaps get closed, and allows 150,000 rules per blocker — five times Chrome’s guarantee. Free: uBlock Origin Lite. $4.99 once: Wipr 2. $9.99 once: AdGuard Pro. $39.99 for lifetime and the most control: 1Blocker. Treat their YouTube claims as this month’s status rather than a feature — each is the developer’s own, and scoped to Safari.
On iPhone and iPad, install a Safari content blocker and stop thinking about browsers — every iOS browser is WebKit underneath. On Android, Firefox for Android with uBlock Origin, because Chrome for Android supports no extensions at all; Google’s own help documents a phone install as “Add to Desktop”.
6. The layer Manifest V3 cannot reach
None of that helps a smart TV, a console or a phone app. DNS filtering does, because it sits outside the browser: point a device or a router at a filtering resolver and requests to known ad domains never resolve. The easy start is NextDNS — free to 300,000 queries a month on unlimited devices, then $1.99 a month. Know one thing first: its own staff say that once you hit the free cap, filtering silently stopsand it behaves as an ordinary resolver until the monthly reset. AdGuard runs free public resolvers needing no account (94.140.14.14 and 94.140.15.15). Quad9 and Cloudflare’s 1.1.1.x variants are not advertising filters at all.
Be clear about the ceiling, because the vendors are. AdGuard’s own DNS documentation says domain-level filtering is coarse: no cosmetic filtering, and it will not remove the whitespace left behind a blocked ad. Nor can it separate a site’s ads from the site itself — Pi-hole’s community FAQ on YouTube concludes in-video ads are not likely to be stopped this way and sends readers to uBlock Origin or YouTube Premium, because Pi-hole sees only domains: blocking YouTube’s ads would mean blocking YouTube. By the same logic it does nothing about “disable your ad blocker” walls, since defusing those needs code inside the page — our reasoning there, not a vendor’s.
To run it yourself, Pi-hole v6.4.3 wants 512 MB of RAM, a static IP and port 53 free, while AdGuard Home v0.107.78 runs natively on Windows and macOS, removing the buy-a-Raspberry-Pi barrier. The install script is the easy part; the work is always-on hardware and the router’s DNS. AdGuard’s own documentation argues against its own upsell here: use DNS filtering in addition to browser filtering, not instead of it. A floor, not a replacement.
7. The ending nobody expected
uBlock Origin proper can never ship on Safari. Apple has never supported blocking webRequest, the API uBO is built on — its README lists Firefox, Chromium browsers and Thunderbird, and that is the whole list. For years the most respected blocker on the web could not exist on the iPhone.
It exists there now. uBlock Origin Lite is on the App Store: free, no in-app purchases, published by Raymond Hill, covering iPhone, iPad, Mac and Vision, updated 24 July 2026, rated 4.72 — the highest of any Safari blocker we checked. The declarative rewrite Chrome forced on him is exactly what made a Raymond Hill blocker possible on the one platform he had been locked out of. Manifest V3 removed the blocking, not the seeing — and the browser that actually delivers the privacy Google claimed for it is Safari.
The ad blockers, ranked by platform
| Tool | Best for | Pricing |
|---|---|---|
1 uBlock Origin | Best on Firefox | Open Source |
| 2uBlock Origin Lite | Best on Chrome | Open Source |
3 AdGuard AdBlocker | Best all-round | Freemium |
4 Wipr 2 | Best one-time purchase (Safari) | Paid |
5 1Blocker | Best lifetime option (Safari) | Freemium |
6 Ghostery | Best open-source alternative | Free |
7 Adblock Plus | Acceptable Ads on by default | Free |
| 8 P Privacy Badger | Best tracker-specific | Free |