For several days, law enforcement has been searching for a man across Italy.
Roadblocks, drones, photos circulated on every available channel. Thousands of reports to be checked one by one.
On Sunday a train was even stopped, but the person they hoped to find was not on board.
This is how an investigation of this scale is conducted in Italy today: scattered, manual, largely dependent on the human factor and on the luck of a chance sighting.
Cases like this reopen the debate on the same unresolved questions surrounding surveillance technology.
Would it really be a problem to have advanced data analysis tools, capable of cross-referencing in real time information that today remains scattered across different databases?
The honest answer is that the problem was never the tool but who controlled it, under what rules and with what scope for independent verification.
In short, among a thousand wrong questions there has always been one right one. The one about governance.

The tool is not neutral, but that is not the point
Palantir, the American company that more than any other embodies this debate, builds platforms capable of integrating public databases, images, communications and administrative records into a single view.
In essence, it collects data wherever data can be collected, then brings it together, cross-referencing millions of pieces of information per second. The perfect surveillance tool, one that raises countless ethical questions. First among them, the conditions under which tools such as Palantir are used: the legal basis, the proportionality between means and end, the real possibility of external oversight of the conduct of those who use them.
And yet it is undeniable that tools like Palantir would help in manhunts such as the one we are witnessing these days. And they would do so with remarkable efficiency, allowing the police to cross-reference CCTV images, phone microphones and phone calls.
It may sound like something out of a 007 film, but this is the (imminent) future.
Whether it becomes a dystopia or a utopia depends solely on how the tool is governed.
What we have and what is missing
The Italian government has set aside a budget for artificial intelligence applied to security, allowing the judiciary to use biometric data screening โin exceptional cases and serious threatsโ, subject to a judge’s authorisation.
Except that the allocated budget (one billion euros) remains far below that of other European countries: 22 billion in the United Kingdom, 10 in France.
The difference also lies, above all, in technological sovereignty: without a European public infrastructure able to compete with American private platforms, every discussion on adopting advanced systems runs up against the same underlying issue: to whom is that data really entrusted?
In Parliament, questions tabled on the matter ask the government whether Italian public bodies already have dealings with Palantir or related companies, and what safeguards exist against the misuse of the most sensitive public data.
It is the right question. Far more so than the recurring but sterile question of whether the technology itself is dangerous or not.

Governance before the tool
A system able to cross-reference mobility data, reports and CCTV footage in real time would probably have sped up a search like the one under way right now. This is a matter of fact, and it would be dishonest to deny it on ideological grounds alone.
But speed is not everything, and it will not be what determines whether technologies like Palantir (if not Palantir itself) are used in the future.
On the scales is the possibility of knowing, always and with certainty, (i) who uses the tool, (ii) for what stated purpose, (iii) with what authorisation, and with what possibility for an independent authority to verify it after the fact.
A data analysis platform without these conditions is a power to be feared.
If, however, all the conditions are met, I believe the idea of trialling its use is the right one.
Every country today faces two options. Two clearly distinct paths.
The first is to import the infrastructure of a foreign private supplier, accepting that sovereignty over its own sensitive data passes through a commercial actor answerable to other jurisdictions. The second is to build, with the necessary investment, a proprietary system. More expensive, but also more secure.
The third option is a non-option, as always. A non-path. That is, not choosing. Doing nothing. And accepting that manhunts like the one of recent days are handled as they were twenty or thirty years ago.
New Connections (FAQ)
If Italy adopted a tool such as Palantir, would it still be under the control of the Italian judiciary?
It would depend entirely on how the agreement was regulated, not on the goodwill of those proposing it. The German case is instructive: in 2023 a court ruled the police’s use of Palantir unconstitutional, yet the tool was not abandoned; instead, the law was amended to make it compatible. This shows that the pressure to use tools already purchased tends to prevail over initial legal objections, unless independent and binding oversight mechanisms are established before adoption, not after.
Why is there so much talk about Palantir and not about other companies that handle vast amounts of data, such as Google or Meta?
Because the nature of the risk is different. Google or Meta accumulate data for commercial purposes, in a context shaped by market forces and competition. Palantir puts its tools directly at the service of state power: intelligence, surveillance, operational decisions by law enforcement. When an advertising platform makes a profiling error, the damage is contained. When the same kind of error is made by a system that steers a police investigation or a national security decision, the consequences bear directly on people’s liberty.
Can a private company that manages sensitive public data refuse to answer a supervisory authority by citing trade secrets?
It is one of the least resolved legal issues, and recent events confirm it: Palantir is known for its extreme secrecy about how its algorithms work, often invoking trade secrets and national security. When a system that shapes public decisions is, by its very contractual nature, opaque even to the institutions that use it, the principle of administrative transparency is hollowed out in practice, even if it remains intact on paper. This is precisely why the most serious proposal circulating among legal scholars is to make any adoption of these tools conditional on an independent, non-negotiable right of audit, before any contract is signed.

