In June, at a media forum in Bamako, a Malian editor named Chahana Takiou stood on a panel with a prosecutor sitting beside him and made a complaint. Mali’s cybercrime law, he said, was being used to jail journalists who should have been tried under the country’s press law, which carries far lighter penalties. Judges, he added, were simply ignoring the press law altogether.
Days later, Takiou was arrested.
Under the cybercrime law.
In August, a court in Bamako sentenced him to a year in prison, half of it suspended, for undermining the credibility of the state. The Committee to Protect Journalists observed, with the weary precision of an organisation that has seen everything, that by arresting him the authorities had proved his comments entirely accurate.
Hold that image for a moment. A man jailed for saying out loud that the law would be used to jail people like him. It is the single most useful thing to keep in your head while Ghana debates its own new law against misinformation.
The apocalypse arrived. Here are the casualties.
For three years we have been warned that generative AI would flood our elections with synthetic lies, that voices would be cloned and faces puppeteered, that democracy itself would drown in fabricated video. The warnings were not wrong. I have written about how AI misinformation exploits trust differently in societies that verify things through people rather than institutions, and I stand by every word of it.
But look at the docket.
In Ghana this year: a TikTok creator in Kumasi arrested in March over a viral video about police officers. A pastor arrested and remanded in April. Two opposition NPP figures detained in April over things they posted online. Another TikToker arrested in May over a video about the President. And in July, Camilla Alhassan, arrested after claiming on TikTok that President Mahama had sacrificed thirty-two cows to secure political power, and that the government’s distribution of sanitary pads to flood victims was a cover for the cows.
She was jailed for a year.
Before sentencing, the court ordered her to take a pregnancy test.
In Uganda last month, Remmy Asiteza of the Daily Express was remanded to Luzira Prison on charges of publishing fake news and broadcasting without a licence, over reporting that alleged corruption in the recruitment of assistant district commissioners. A second Ugandan journalist had already spent five weeks inside on a fake news charge over a report about the First Lady. Those charges were dropped in August, after the five weeks.
Now count the deepfakes in that list.
Zero. Not one synthetic voice. Not one manipulated frame. No generative anything. Just people saying things. Some of it absurd, some careless, some inconveniently accurate. All of it prosecuted under laws that were sold to the public in the vocabulary of a technology panic.
The law is written for the robot and used on the neighbour
This is the trick, and it is not subtle once you see it.
“We need this law because of deepfakes” is an extraordinarily easy sell. Deepfakes are frightening, foreign, technical and unanswerable. Nobody marches in defence of synthetic media. The alternative pitch, “we need this law because criticism of the governing party keeps embarrassing us,” polls considerably worse, even when it is the more accurate description of how the law behaves once it exists.
So, you build the infrastructure for the robot. And then you discover that the infrastructure cannot distinguish between a fabricated video engineered to trigger a riot and a blogger who got a fact wrong about a minister’s contract. Both are, technically, false information published online. The statute cannot tell them apart. The statute was never asked to.
Professor Stephen Kwaku Asare put it precisely last month. Section 208, he wrote, “invites the police to become the immediate arbiters of truth.” Arrest, he added, should not be the state’s first response to a disputed claim.
Ghana’s numbers make the point with uncomfortable clarity. The Media Foundation for West Africa counted fourteen arrests under Section 76 of the Electronic Communications Act and Sections 207 and 208 of the Criminal Offences Act in sixteen months under the current administration. Across the entire eight years of the administration before it, the figure was eight.
Before anyone reaches for a party card: in July the Minority Caucus demanded that the President direct the Attorney-General to repeal Sections 207 and 208, with Alexander Afenyo-Markin describing Alhassan’s imprisonment as a recreation of criminal libel. He is right. He is also leading a party that used the same sections for eight years, and whose own proudest claim is that it abolished criminal libel in 2001 and then left these two provisions standing. That is the tell. A law that only one side finds useful is a bad law. A law that every side discovers a fondness for the moment it takes office is a dangerous one.
The cure is already on the order paper
Which brings us to the part that should worry Ghanaians most, and which has attracted a fraction of the attention the arrests have.
Sitting in the legislative pipeline, part of a package of digital governance bills from the Ministry of Communications, Digitalisation and Innovation, is the Misinformation, Disinformation, Hate Speech and Publication of Other Information Bill. It proposes to repeal Section 208 and amend Section 76, which is exactly what the Media Foundation for West Africa, the Ghana Journalists Association and now the Minority have all been demanding.
Read that again and then read the bill.
The MFWA, which has campaigned for the repeal of Section 208 for years, has raised the alarm about what would replace it. The bill contemplates a licensing-based sanction regime covering media and content creators, which runs into Article 162(3) of the Constitution, the clause that forbids censorship of the media through licensing. It defines hate speech expansively enough to include promoting negative feelings, a threshold low enough to capture most of Ghanaian political commentary and a good deal of Ghanaian comedy. And it keeps custodial sentences for speech offences, which is how you rebuild criminal libel while holding a press conference about repealing it.
The GJA has rejected the bill in its present form. The President has said the bills are not designed to restrict free expression, and I have no reason to doubt that he means it. The problem is that intention is not the operative part of a statute. Section 208 was not drafted by people hoping to jail a woman over cows. It was drafted in 1960, inherited from colonial administrators who had their own reasons, and it sat there patiently until someone found a use for it.
Note also what Ghana is doing in parallel. A new Data Protection Bill is being prepared to bring AI systems, automated decision-making and deepfakes under formal oversight, with a separate Emerging Technologies Bill behind it. Which means the actual AI problem, the one we were told justified all of this, is being addressed somewhere else entirely, by regulators, in the language of data and systems.
So, what exactly is the speech bill for?
The cases are not clean, and pretending otherwise is a mistake
Let me concede what a lazier version of this argument would skip.
Some of these cases are genuinely ugly. Pastor William Gyimah was charged with making death threats against the Vice President, not with criticising her. Threats are not journalism and nobody should pretend they are. Alhassan’s cow story was not investigative reporting. It was nonsense, and it was nonsense that travelled.
That is exactly the problem. The hard cases are where these powers get their legitimacy.
Nobody builds a speech-policing apparatus by starting with the reporter they want to silence!
You start with the death threat, the obvious lunatic, the video everyone agrees is beyond the pale. You establish that the power is reasonable, necessary, wielded by serious people against unserious ones.
And then, sixteen months later, it reaches the person covering the procurement scandal.
The question was never whether some of the people arrested deserved consequences. It is who holds the pen that defines the category, and what happens when the person holding it is also the subject of the story.
What a serious version would require
I came into this research from a stint in the newsroom, and I left carrying an unglamorous conclusion. The loudest voices demanding protection from misinformation are not reliably the honest ones. Sometimes the rot is in the ministry. Sometimes it is in the media house. Very often it is in the relationship between them, which is precisely why neither should referee the other alone.
If the MDHI Bill is to be worth passing, three things have to change in it.
Strip out the custodial sentences. A fine, a correction order, a civil remedy. Prison for speech is how you convert a regulatory instrument into a political one, and Ghana already repealed criminal libel once. Doing it twice would be embarrassing.
Narrow the definitions until they bite only on real harm. Promoting negative feelings is not a legal standard, it is a mood. A synthetic video engineered to incite violence during an election and an insulting remark about a minister are not the same offence, and any statute that cannot say so in plain language is not fighting misinformation. It is fighting scrutiny, and has borrowed misinformation’s better press.
Keep the licensing clauses out, and put the regulator beyond the Executive’s reach. Article 162(3) exists for a reason. A body deciding what counts as false cannot be staffed by appointees of the government most likely to be the subject of the falsity claim, and journalists need to be in the drafting room rather than in the press release.
The cost nobody is pricing
Here is what worries me most, and it is not the arrests.
Every time a reporter is jailed on a fake news charge, the public learns something. Not that the story was false. Ghanaians are not stupid, and can see perfectly well which stories attract prosecutions and which do not. What the public learns is that “fake news” is a political instrument rather than a factual category.
Watch what happens next door. South Africa goes to local elections on 4 November under a new Disinformation Code requiring parties to label AI-generated material and correct false claims within thirty-six hours. Days before the electoral commission told parties to label synthetic content, it posted an AI-altered image of its own to promote voter registration. Nobody flagged it. These regimes are only ever as credible as the institution running them.
So, consider where this leaves us. In two years, or five, a genuinely sophisticated deepfake will circulate in Ghana on the eve of something that matters. The government will step forward and say: this is fabricated, this is synthetic, do not believe it.
And they will be telling the truth. And nobody will believe them.
We are discrediting the antibody before the infection has fully arrived. The people who will pay for that are not the ministers currently enjoying the convenience of Section 208. It is everyone who, in some future emergency, needed there to be one institution left in this country whose word still meant something.
The question was never how we stop fake news. It is who gets to decide what’s fake, and whether that person has any incentive whatsoever to be honest about it.
Abugre Alebsuure Abayeta is a doctoral researcher in intercultural communication at the University of Jyväskylä, Finland, working on algorithmic bias and intercultural digital inclusion in generative AI.
