A year and a half after the reform, Hesse has declared its new policing legislation a success. The figures presented do indeed show that the new tools are being used extensively. However, it is not yet possible to determine from them whether they have made the state measurably safer.

“The new policing legislation is working and has made Hesse even safer.” With this statement, Hesse’s Minister for the Interior, Roman Poseck, draws a remarkably clear conclusion regarding the amendment to the Hessian Act on Public Safety and Order (HSOG), which was adopted at the end of 2024.

At first glance, the figures appear impressive: 521 recorded breaches of relevant police orders, 69 instances of electronic tagging used for risk prevention, more than 3,400 orders for preventive detention, 772 bodycam deployments in homes and 28 requests for AI-assisted video searches.

There is therefore no question that these new tools are being used. Yet this is precisely where the problem with the presented track record lies: use does not in itself prove effectiveness.

Anyone claiming that a legislative reform has made a federal state safer must demonstrate more than just the number of times the measures have been applied.

3,400 applications do not mean that danger was averted 3,400 times

This becomes particularly clear in the case of preventive detention. According to the Ministry of the Interior, this measure has been ordered more than 3,400 times since the reform, of which over 1,950 cases were for a duration of up to six days.

This is a significant figure – precisely because it constitutes a particularly far-reaching state intervention. People can be deprived of their liberty before the offence that is to be prevented has even been committed.

This may be necessary in situations of concrete danger. A state governed by the rule of law must give security authorities the means to prevent a serious offence that is an imminent threat, rather than waiting until it actually occurs.

It is precisely for this reason, however, that a more detailed evaluation would be necessary.

How many serious offences have been demonstrably prevented through preventive detention? In which situations of imminent danger was it used? How often was the maximum possible duration actually utilised? How did the cases develop after the measure was lifted? And how often were such decisions reviewed or amended by the courts?

The claim that more than 3,400 orders demonstrate the success of the reform falls short. A high number can just as easily indicate that a tool is needed and is being used consistently. However, taken on its own, it says nothing about how targeted and effective it is.

Electronic tagging: potential with a specific protection interest

The situation is different with electronic tagging, which has been used 69 times to avert danger since the reform came into force – according to the Ministry of the Interior, frequently to protect women from domestic violence.

Here, the security benefits are particularly clear. Where a specific threat exists and technical monitoring can help to better protect those at risk, electronic tagging can be an important tool within a comprehensive protection strategy.

The fact that the number of cases nearly doubled within a further six months, following 35 cases in the first year, also shows that the tool is being used increasingly in practice.

But here too, the key metric should not be how many ankle monitors have been ordered. What matters is what has been achieved as a result: How many attempts to approach the victim were detected? How often were the emergency services able to respond in time? Did further assaults occur despite the measure? And how do those affected and specialist agencies assess the actual level of protection?

A tool designed specifically to protect people in concrete situations of threat particularly deserves an evaluation that measures its success in terms of its protective effect rather than the frequency of its use.

AI-assisted video surveillance: Individual cases are convincing – but do not replace an overall assessment

The Ministry of the Interior presents a particularly clear case for AI-assisted video surveillance. In 28 real-life cases, targeted searches have so far been requested, predominantly to locate missing persons, including 18 minors.

The case of a missing 16-year-old girl, who, according to the Ministry, was in an acute crisis situation and was located and brought to safety with the help of the system, demonstrates the concrete benefits that intelligent video analysis can offer.

This should not be downplayed. If technology helps to find a person at risk more quickly, that is a strong argument for its targeted use.

However, one successful case does not answer all the questions associated with AI-supported video surveillance.

What is the hit rate? How many false alarms are generated? What criteria may be used for searches? What data is processed, for how long is it stored, and who can access it? How can we prevent a clearly defined tool from gradually evolving into an ever more comprehensive automated surveillance infrastructure?

When it comes to AI in particular, the following principle must apply: the more powerful the technology, the more transparency, oversight and clear boundaries must keep pace with its development.

Bodycams in homes encroach upon a particularly protected space

The expanded use of bodycams must be viewed with a similar degree of nuance. Since the reform, they have been deployed in homes 772 times to avert, according to the Ministry, imminent threats to life, limb or liberty.

For police officers, bodycams can offer protection in escalating situations, document events and potentially help to de-escalate them. Operations in homes, in particular, can be highly dynamic and dangerous.

At the same time, the home is a particularly protected private space. It is therefore right to impose particularly high standards on operations carried out there.

Here, too, the figure of 772 instances is not sufficient proof of success. It would be far more interesting to know whether attacks on officers have decreased, whether recordings have helped to clarify contentious operational situations, and in which situations the cameras actually improved safety.

Frankfurt’s station district highlights the limitations of pure statistics

Of the 521 recorded breaches of police orders, 478 relate to exclusion orders, according to the Ministry. 405 breaches were recorded in the Frankfurt am Main Police Headquarters area alone, particularly in the station district.

Poseck argues that exclusion orders prevent, for example, drug dealers from ‘returning time and again to the scene of the crime and, in a sense, making a mockery of the security authorities’.

The approach is understandable: anyone who repeatedly commits offences or poses a specific threat in a particular location should not be able to return there immediately.

However, the figures also raise another question. If hundreds of breaches are recorded despite existing bans, to what extent does this measure bring about a lasting change in the security situation?

Are crime and problematic structures actually being reduced – or are they, in some cases, merely being shifted to other locations? Is the public’s sense of security improving? Is the number of relevant offences in the affected area falling in the long term?

In the Frankfurt railway station district in particular, the impact of the new police legislation should therefore not be measured solely by the number of exclusion orders issued and breached, but by the actual development of the security situation.

Security and freedom are not an either/or situation

The new Hessian police legislation contains measures for which there are comprehensible security policy reasons. Protecting people from domestic violence, searching for minors at risk, safeguarding Jewish institutions, or preventing specific serious crimes are tasks that require a state capable of taking effective action.

A critical examination of the HSOG therefore does not mean fundamentally denying the police the use of modern tools.

On the contrary: security authorities must be able to keep pace with technological and societal changes. New threats may require new responses, as Poseck notes.

But new responses also require new forms of oversight.

The more preventive measures encroach on civil liberties and the more powerful digital surveillance and analysis tools become, the more important proportionality, transparency, data protection, parliamentary oversight and independent evaluation become.

A track record requires impact data

After one and a half years, it is legitimate to take stock. The figures presented do reveal something important: the powers created by the HSOG amendment have not remained mere theoretical instruments. They are being used in practice.

That is a relevant observation.

However, it is not sufficient to support the broader claim that the new police legislation has ‘made Hesse even safer’.

A convincing assessment would need to combine operational statistics with impact data: crimes prevented, trends in relevant areas of crime, the protective effect in cases of domestic violence, the outcomes of preventive detention, hit and error rates of AI-supported systems, the impact of bodycams, and judicial reviews of the new powers.

Security policy should not be judged by how many tools the state deploys, but by whether they achieve what they were designed to do – and whether this is done proportionately.

The first year and a half of the new HSOG therefore provide good grounds for an interim assessment.

It is still too early to declare it a ‘complete success’.

[ML]