France's Council of Ministers examined a bill on July 9, 2026 that would substantially expand the power of PHAROS, the state's online-reporting platform, to compel takedowns of racist and antisemitic content. Presented by Aurore Bergé, the minister delegated for the fight against discrimination, the projet de loi de cohésion républicaine (PJL 25-890) is scheduled for accelerated procedure, with Senate examination beginning in October 2026.
What the Bill Actually Does
PHAROS was created to flag child sexual abuse material and terrorism apology — categories where illegality is, in the overwhelming majority of cases, unambiguous on its face. Article 7 of the new bill extends that same removal mechanism to racist and antisemitic content. Once PHAROS flags material as "manifestly illicit," platforms must remove, delist, or block access to it, with internet service providers on the hook to block non-compliant sites entirely, according to the bill's provisions as reported by Le Club des Juristes. The government's own explanatory memorandum cites a more than 45% rise in PHAROS content reports since 2023 as the justification for scaling up the mechanism (Sénat, exposé des motifs, PJL 25-890).
The bill also creates a mandatory supplementary ineligibility penalty for elected officials convicted of hate-motivated felonies carrying at least three years' imprisonment, and a discretionary ineligibility penalty for denial of crimes against humanity — narrower than an earlier version the Constitutional Council flagged for potentially covering speech-only offenses (Le Club des Juristes). It further ends a long-standing press-law exemption by making organizations, not just individuals, criminally liable for hateful publications, and lets the state file complaints directly on behalf of public employees targeted by racist or antisemitic abuse (Public Sénat).
The bill is a deliberately narrowed successor to the Yadan bill, an earlier antisemitism-focused proposal that was withdrawn in April 2026 after triggering a 700,000-signature petition and warnings that its language on "implicit" incitement and state-destruction rhetoric would sweep in ordinary political speech about Israel and Palestine. Bergé's version drops the most contested Yadan provisions and broadens the frame to racism generally, aiming for cross-party consensus rather than a repeat fight.
The Case For It
The strongest argument for the bill is numerical, not rhetorical. French police and gendarmerie recorded more than 16,000 racist, xenophobic, or antireligious offenses in 2024 — 9,400 crimes and délits, an 11% year-on-year increase — with the majority consisting of insults, provocations, and defamation, much of it migrating online (DILCRAH). PHAROS reports have outpaced that growth. A state channel that can get genuinely illegal racist and antisemitic content off major platforms quickly, rather than waiting on a courtroom process that can take months, addresses a real and growing harm, and France is not inventing the concept: it already has notice-and-blocking powers for terrorism apology and CSAM that have not, by most accounts, produced runaway censorship. Treating racist and antisemitic incitement with comparable urgency is a defensible policy judgment on its own terms.
Why the Mechanism Is the Wrong Fix
The flaw is not the goal but the method. CSAM and terrorism apology are categories where an administrative reviewer can apply an essentially binary test — is this the thing, or not. Racist and antisemitic speech is not that. It routinely turns on context, intent, satire, and historical or political framing that a 24-hour administrative removal cycle is poorly built to weigh. The bill's own legal reviewers have flagged this: Le Club des Juristes' summary of the text notes explicit concern that "manifestly" illicit and discriminatory-intent standards remain "malleable" and dependent on "administrative discretion" — precisely the vagueness that due-process safeguards in speech law exist to prevent.
Compressing that judgment call into an administrative flag-and-remove pipeline, enforceable by ISP-level blocking, removes the judge from the loop at the exact moment judicial line-drawing matters most. It also hands platforms a strong incentive to over-comply: facing blocking orders for resisting a government flag, the economically rational move is to take content down first and let disputes sort themselves out later, if ever. That dynamic is not unique to France — the European Court of Justice's Coyote System ruling this July already pushed EU platform-liability law toward broader intermediary exposure, with the EFF warning that "freedom of expression online" risks becoming "collateral damage" as courts and legislators keep expanding what counts as platform "control" over content (EFF). France's PHAROS expansion fits that broader European pattern of shifting adjudication away from courts and onto platforms under compressed deadlines.
None of this requires abandoning the bill's underlying aim. A takedown regime that preserved judicial referral for contested cases — reserving the fast administrative track for content that is unambiguous, as PHAROS's original CSAM and terrorism mandate was — would capture most of the stated benefit without the vagueness problem. As currently drafted, the bill asks an administrative flag to do a judge's job on a 24-hour clock, for a category of speech that resists 24-hour clocks by its nature. The Senate's October 2026 review is the moment to fix that design choice, not just debate whether the underlying goal is worthy.