Sofia: Bulgaria’s prosecutors will soon have to charge a category of conduct their own criminal code barely recognises, because the EU anti-corruption directive that the Council adopted on 21 April 2026 obliges every member state to criminalise the abuse of functions by public officials. That single offence held up the negotiation for months, and it is now law that capitals must copy into national statute books.
The directive is the Union’s first attempt to write one criminal standard for corruption across twenty-seven legal traditions. Parliament and Council reached their provisional deal on 2 December 2025, the civil liberties committee endorsed it on 27 January 2026 by 54 votes to 11 with nine abstentions, and the plenary signed off in March. The legislative file ran for nearly three years, which tells you how sensitive the subject remains.
What the text actually does is narrow. It sets common definitions for bribery, embezzlement, trading in influence, obstruction of justice and illicit enrichment. It harmonises penalty ranges so that the same conduct does not draw a suspended sentence in one country and a decade in another. It requires governments to run prevention work rather than only prosecution, and it mandates annual publication of EU-wide corruption data, which is the first time anyone will be able to compare enforcement across the bloc on a common basis.
The abuse of functions offence is the piece that matters most and travels worst. Several governments argued during the trilogues that criminalising an official’s misuse of discretion risks pulling ordinary administrative error into criminal court, and that prosecutors in politically contested systems could use it against opponents. Anti-corruption campaigners argued the opposite, that without it the most damaging conduct in procurement and licensing stays outside the criminal law entirely. The compromise text keeps the offence and leaves capitals to draw the boundary.
That is where the directive will succeed or fail. A directive binds governments to a result and lets them choose the means, so twenty-seven parliaments will now produce twenty-seven versions of the same offence. Some will draft it tightly. Others will draft it so narrowly that it catches almost nothing, and the Commission will have to decide whether to open infringement proceedings over the wording of a criminal statute, which it has historically been reluctant to do.
The Commission’s own diagnostic work already shows why the gap should worry people. The 2026 Rule of Law Report, published on 17 July, assesses anti-corruption frameworks alongside judicial independence, media pluralism and institutional checks in all member states and four candidate countries. Its recurring finding is that laws exist and enforcement lags, particularly in high-level cases where the defendant has political cover.
A criminal directive does not fix that. It gives national prosecutors a clearer legal basis and gives Brussels a comparable dataset, and both are real gains. Neither supplies investigators, court time or the political tolerance for prosecuting people in office. Bulgaria, Romania, Hungary and Slovakia have each spent years demonstrating that the binding constraint is rarely the statute.
Still, the annual data obligation may end up doing quiet work. Once every government reports convictions, case durations and sanctions on a common template, the ones that report almost nothing will stand out, and that comparison will land in the middle of the next budget negotiation. Conditionality arguments are far easier to make with numbers than with narrative country chapters.
