AI and Accessibility for People with Disabilities at INSS

This page is a translation of the original Portuguese text, provided for convenience. It deals with Brazilian law. If there is any discrepancy, the Portuguese version prevails. Read the original in Portuguese

Artificial intelligence is transforming how people with disabilities access benefits from INSS, but technological advances still coexist with barriers that call for legal and social attention.

The Current State of Digital Accessibility at INSS

When you look at the relationship between technology and social security rights in Brazil, one thing stands out: INSS has gone through a rapid digital transformation in recent years. Meu INSS (the agency’s online portal and app), the central tool for filing claims and tracking benefits, has become the main service channel for millions of insured persons. For people with disabilities, this shift to digital is a double-edged sword: it removes the physical barrier of getting to an INSS office, but it also creates new challenges of digital accessibility.

People with visual, hearing, motor, or cognitive disabilities face specific obstacles when they use digital systems. Screen readers do not always interpret the Meu INSS interfaces correctly, complex forms make voice-command navigation difficult, and the technical language of social security becomes even more exclusionary when it comes through a screen. Despite institutional efforts, the platform still has significant gaps in compliance with the Web Content Accessibility Guidelines (WCAG), the international standard that sets criteria for making digital content usable by everyone.

Articles 203 and 227 of the Federal Constitution guarantee special protection to people with disabilities. The Brazilian Inclusion Law (Law 13.146/2015) reinforces the right to accessibility in digital environments, including public services. The Statute of Persons with Disabilities provides that the government must promote assistive technology, which includes adapting government platforms. In that context, building artificial intelligence into INSS systems is not merely a matter of modernization. It is a matter of meeting legal obligations.

How Artificial Intelligence Can Expand Access to Benefits

Artificial intelligence offers concrete ways to reduce the barriers that people with disabilities face when they seek social security services. Chatbots with natural language processing can simplify service by translating technical terms into plain language and walking the insured person through the claims process in a conversational way. Voice recognition systems allow people with motor disabilities to use the platform without typing. Automatic image description tools help people with visual disabilities understand the documents and proof that INSS requires.

One of the most promising uses of AI in the social security context is automated document screening. People with disabilities often need to submit medical reports, certificates, and statements to prove their condition. Intelligent systems can read these documents using optical character recognition (OCR), extract the relevant information, and pre-fill forms, significantly reducing the effort required of the claimant. This kind of automation is especially relevant for benefits such as the BPC/LOAS (the Continuous Cash Benefit, a monthly assistance benefit for people with disabilities and older adults in need), which requires proof of both disability and socioeconomic status.

Another important area is the medical examination. INSS already uses document-analysis tools for some disability-related benefits, and expanding that technology to review reports of people with disabilities could speed up processes that now take months. Trained algorithms can spot patterns in medical records, suggest classifications, and flag cases that need a more detailed in-person evaluation, prioritizing cases by urgency.

True digital inclusion at INSS is not just about putting services online. It is about making sure every person, whatever their condition, can exercise their social security rights with autonomy and dignity.

AI also has potential for personalizing service. Intelligent systems can identify the insured person’s profile (type of disability, history of interactions, benefits already requested) and adjust the interface automatically: enlarging fonts, turning on high contrast, offering audio description, or simplifying the navigation flow. This kind of dynamic personalization goes beyond manual accessibility settings by anticipating needs based on the user’s own data.

Risks and Challenges: When Technology Reproduces Exclusion

Despite its transformative potential, applying artificial intelligence to social security systems carries risks we cannot ignore. The first is algorithmic bias. AI systems are trained on historical data, which may contain discriminatory patterns. If the data used to train a screening model reflects past decisions that disproportionately denied benefits to certain groups, the algorithm tends to reproduce that inequality. For people with disabilities, this can mean automated denials based on criteria that do not adequately account for the diversity of conditions and their functional impact.

The digital divide also remains a structural obstacle. Data from IBGE (Brazil’s national statistics bureau) show that people with disabilities have less access to the internet and to technological devices than the general population. Making INSS services digital by default, without keeping effective alternative channels, can deepen that inequality. In that scenario, AI risks becoming one more barrier instead of a bridge.

Privacy and data protection deserve extra attention when it comes to people with disabilities. Information about health conditions is sensitive data under the LGPD (Brazil’s General Data Protection Law). Using AI to process medical reports, clinical histories, and functional assessments calls for robust safeguards. Any leak or misuse of this information can lead to discrimination in other areas of the insured person’s life, such as the job market or the purchase of private insurance.

Another significant challenge is the transparency of automated decisions. When an AI system takes part in reviewing a benefit claim, the insured person has the right to understand the criteria behind the decision. The so-called “black box” of algorithms, where the decision-making process is opaque even to the developers themselves, conflicts with fundamental principles of administrative law, such as the duty to state reasons for official acts and the right to be heard and to respond. For people with disabilities, who already struggle to understand bureaucratic decisions, algorithmic opacity adds another layer of exclusion.

The Role of Law in Regulating AI in Social Security

Brazil’s legal system already provides a solid basis for requiring that AI in INSS services respect the rights of people with disabilities. The Brazilian Inclusion Law (the Statute of Persons with Disabilities) requires the government to ensure accessibility across all of its digital platforms. The LGPD establishes the right to have automated decisions reviewed and to have the criteria used explained. The AI Legal Framework (once fully regulated) is expected to bring specific guidelines on the use of artificial intelligence in public services.

Social security lawyers have a key role to play here. Specialized attorneys need to understand the AI mechanisms INSS uses so they can identify possible irregularities in automated decisions, challenge denials based on algorithmic criteria, and make sure the right to human review is actually honored. The intersection of social security law and technology requires legal professionals to keep their knowledge constantly up to date.

At the international level, the Convention on the Rights of Persons with Disabilities (ratified by Brazil with the status of a constitutional amendment) requires states to promote access to new information and communication technologies. This international commitment reinforces INSS’s obligation to make sure its technology solutions are inclusive by design (the principle of universal design), and not merely adapted afterward as a stopgap.

Oversight of INSS’s digital accessibility can take several forms: public civil actions brought by the Public Prosecutor’s Office, complaints to the Federal Court of Accounts (which has already addressed accessibility on government platforms), reports to the National Data Protection Authority, and individual lawsuits by insured persons whose rights were harmed by a lack of accessibility.

Paths to a Truly Inclusive Digital Social Security System

For artificial intelligence to become a real ally of people with disabilities in accessing social security benefits, we see several essential measures. First, the development of any AI solution for INSS must involve people with disabilities from the earliest design stages. Usability testing with real users, accessible public consultations, and the participation of representative organizations are indispensable to ensure that the technology meets this community’s actual needs.

Training INSS staff to work with assistive technologies and to meet the specific needs of people with disabilities is just as important. AI should support human service, not replace it. Complex cases, involving multiple disabilities or delicate socioeconomic circumstances, call for a sensitivity that no algorithm can adequately reproduce.

We consider it essential to create a hybrid service channel that combines the advantages of intelligent automation with the option of human intervention at any point. An insured person with a disability should be able to move from automated service to in-person service, or to a videoconference with a Libras (Brazilian Sign Language) interpreter, without losing the progress already made in the system. This integration across channels is technically feasible and legally necessary.

Finally, algorithmic transparency must be treated as a priority. INSS should publish periodic reports on how its AI systems perform, including accessibility metrics, error rates by user profile, and the impact of automated decisions on different groups of insured persons. This kind of accountability allows civil society, the Public Prosecutor’s Office, and oversight bodies to track whether the technology is fulfilling its inclusive purpose or deepening inequality.

As informações deste artigo são de caráter informativo e não substituem consulta jurídica individualizada. Cada caso possui particularidades que exigem análise específica.