Can AI Solve Latin America's Crime Crisis?
Brazil's Pax Is Betting That Crime Fighting Is a Data Problem
Latin America has long faced a difficult paradox. It is home to some of the world's most dynamic economies, youngest populations, and fastest-growing technology sectors, yet it also experiences some of the highest levels of violent crime. For decades, governments have poured resources into policing, tougher sentencing laws, and security operations, often with mixed results. The challenge has never been a lack of effort. It has been a lack of information.
Now, a Brazilian startup believes artificial intelligence may help change that equation.
Pax, a São Paulo-based public safety technology company, recently raised a $40 million seed round—one of the largest seed investments in Latin American history. The funding was co-led by Greenoaks and Benchmark, investors known for backing companies such as Uber, Instagram, and Anthropic.
The investment reflects growing confidence that AI can do more than improve office productivity. Increasingly, investors and governments are exploring whether artificial intelligence can address some of society's most complex challenges—including crime.
But can AI genuinely help solve Latin America's public security crisis, or does it risk creating new challenges around surveillance, privacy, and civil liberties?
The Data Problem Behind Crime
The modern criminal investigation often suffers from an overwhelming volume of disconnected information.
Police departments typically manage thousands of surveillance cameras, vehicle registration databases, criminal records, emergency reports, witness statements, and intelligence files. Much of this information exists in separate systems that rarely communicate with one another.
As a result, investigators spend significant time manually connecting pieces of information that already exist.
Pax's core argument is that crime solving is fundamentally a data integration problem.
The platform connects fragmented camera networks, criminal databases, vehicle records, and geospatial information into what the company calls a "live intelligence graph." Artificial intelligence algorithms then identify patterns, generate investigative leads, and alert officers to connections that might otherwise go unnoticed.
According to founder and CEO David Peixoto:
"Solving crimes is a data problem. The officer investigates. The officer decides. The platform multiplies their capacity."
This distinction is important.
The company's objective is not to automate policing decisions. Rather, it seeks to enhance human investigators by reducing the time required to identify suspects, track movements, and establish links between incidents.
Early Results Are Attracting Attention
Pax claims that deployments have already contributed to:
- A 27% reduction in violent crime over six months in certain deployment areas
- A doubling of investigative efficiency
- More than 2,000 criminal cases resolved through platform-supported investigations
The platform has reportedly assisted investigations involving:
- Homicides
- Armed robberies
- Vehicle theft
- Phone theft
The latter is particularly significant in Brazil, where smartphone theft affects millions of citizens annually and has become one of the country's most common crimes.
If these figures prove sustainable at scale, they could represent one of the most significant examples of AI delivering measurable public safety outcomes.
Why Venture Capital Is Paying Attention
For years, venture capital largely focused on enterprise software, fintech, and consumer technology.
AI is beginning to shift those priorities.
The Pax investment signals a growing belief that artificial intelligence can become a foundational layer of public infrastructure rather than merely a productivity tool.
Public safety, transportation, healthcare, and critical infrastructure are increasingly viewed as sectors where AI may generate both social impact and significant economic returns.
Greenoaks partner Andrew Cohen summarized the opportunity succinctly when noting that asking Brazilians what issue they most want fixed often produces the same answer: crime.
From an investor's perspective, solving even a small portion of that challenge represents a massive market opportunity.
The Promise of AI-Enabled Public Safety
The appeal of AI in law enforcement is straightforward.
Criminal activity increasingly generates digital footprints through mobile devices, cameras, financial transactions, vehicle movements, and online communications.
Humans struggle to process such enormous volumes of information in real time.
Artificial intelligence excels at identifying patterns across large datasets.
Potential advantages include:
Faster Investigations
AI can significantly reduce the time required to analyze surveillance footage, identify vehicles, or cross-reference suspects across multiple databases.
Better Resource Allocation
Police departments often operate with limited personnel. AI systems can help prioritize cases, identify crime hotspots, and focus investigative resources where they are most needed.
Improved Crime Prevention
Rather than reacting after crimes occur, AI may help identify emerging patterns and provide early warning indicators.
Enhanced Situational Awareness
Integrated intelligence systems can give investigators a more complete operational picture, improving decision-making and inter-agency coordination.
For regions facing persistent security challenges, these capabilities could provide meaningful operational advantages.
The Risks Are Equally Significant
However, deploying AI in public safety environments introduces complex ethical and governance challenges.
The same technologies that help solve crimes can also expand surveillance capabilities.
History demonstrates that powerful security tools often generate concerns about misuse, discrimination, and accountability.
Several risks deserve particular attention.
Privacy and Civil Liberties
Large-scale integration of camera networks and personal data raises legitimate concerns regarding individual privacy rights.
Without strong legal safeguards, such systems could evolve into mechanisms for mass surveillance rather than targeted law enforcement.
Algorithmic Bias
AI systems learn from historical data.
If historical policing data reflects existing biases, algorithms may inadvertently reinforce unequal treatment of certain communities.
Transparency and Accountability
Citizens have a right to understand how decisions affecting public safety are made.
Opaque algorithms can create accountability gaps if authorities cannot explain how investigative leads were generated.
Function Creep
Technologies introduced for serious crime investigations may gradually expand into broader monitoring activities beyond their original purpose.
The challenge is ensuring that technological capability does not outpace democratic oversight.
Governance Will Determine Success
The long-term success of AI-enabled policing will depend less on technological sophistication and more on governance.
Effective frameworks should include:
- Independent oversight mechanisms
- Clear legal authorities governing data use
- Audit trails for all system queries and investigations
- Human review requirements for operational decisions
- Regular assessments of accuracy and bias
- Transparent reporting to the public
Notably, Pax emphasizes that all system queries are logged and auditable—a feature increasingly viewed as essential for responsible AI deployment.
Trust will be critical.
Without public confidence, even highly effective systems may face resistance.
A Preview of the Future?
The rise of Pax may represent a broader trend in how governments approach security challenges.
Artificial intelligence is rapidly moving from experimental pilot projects to operational deployment across public institutions.
Just as AI is transforming finance, healthcare, logistics, and cybersecurity, it is beginning to reshape public safety.
The question is no longer whether AI will enter policing.
The question is how societies ensure that its deployment enhances security while preserving democratic values.
Conclusion
Artificial intelligence alone will not solve Latin America's crime crisis.
Crime is ultimately rooted in social, economic, political, and institutional factors that technology cannot eliminate.
However, AI may significantly improve the ability of law enforcement agencies to investigate crimes, allocate resources, and identify threats.
The experience of Pax suggests that when properly governed, AI can function as a force multiplier rather than a replacement for human judgment.
The real challenge lies in balancing effectiveness with accountability.
If Latin America can achieve that balance, AI may become one of the most important public safety tools of the coming decade.
If it cannot, the region risks trading one security challenge for another.
The future of AI in policing will therefore be determined not only by what the technology can do, but by how responsibly societies choose to use it.
