Seven markets. One binding AI rulebook for lenders, and it is not Brazil's.

Oonagh van den Berg | RAW Compliance

Issue 4 mapped Europe, where one AI Act met a long tail of national supervisors. Latin America has the opposite shape. Most of the region is still debating bills and publishing central bank warnings. One country has already written credit scoring into its AI rules.

That country is Peru. Brazil, the region's largest financial market, has a bill that has not reached a Chamber vote and a central bank that says it is studying the topic. A bank cannot treat "LATAM" as one jurisdiction. It has one market with live duties, several with pending bills, and a few where the honest answer is that nothing binding has been found.

Peru - the reglamento that names credit scoring

Peru's Regulation of Law 31814 was approved by Supreme Decree 115-2025-PCM and published in El Peruano on 9 September 2025. It takes a risk-based approach. Among the uses it treats as high risk is determining the credit assessment of persons, unless the AI is used to detect financial fraud. It also lists AI used to manage critical assets that support essential services, banking among them. Developers and deployers of high-risk systems must provide algorithmic transparency under Article 25, which includes explaining the criteria behind automated decisions that affect people's rights. The decree enters into force 90 business days after publication, apart from four final provisions that took effect the next day. BDO Peru reports that the first compliance deadline for private developers and deployers in six sectors, including economy and finance, fell on 10 September 2026, with later dates for other sectors. A lender using AI to score customers in Peru should read this as a live rule, not a consultation.

Sources: https://busquedas.elperuano.pe/dispositivo/NL/2436426-1 ; https://www.gob.pe/institucion/pcm/normas-legales/7133522-115-2025-pcm ; https://www.bdo.com.pe/es-pe/publicaciones/consultoria-de-negocios/2026/del-uso-de-ia-a-su-gobernanza-claves-para-llegar-preparados-al-10-de-setiembre

Brazil - a bill in the Chamber and a central bank that wants the sector role

PL 2338/2023, the AI legal framework, was approved by the Senate in December 2024 and sits in a special committee of the Chamber of Deputies. Trackers cited by CiberLATAM say the committee has no report and there is no official vote date. As drafted, it would create a national AI regulation and governance system coordinated by the data protection authority, ANPD, with the National Monetary Council and the Central Bank as sector regulators for finance. On 1 September 2026, a coordinator in the Central Bank's financial system regulation department said at a Brasilia event, as reported by Valor, that Brazil has no AI-specific regulation, that the Bank sees itself as the sector regulator if the framework passes, and that it is worried about concentration among AI service providers used by regulated firms. He also said the Bank is looking at how risk-modelling practices absorb AI risks. In March 2026 the same official told another event that the Bank was studying AI regulation. Separately, Joint Resolution 18 of the CMN and the Central Bank, in force from 1 January 2026, requires a data quality policy with traceability and named accountability, with a final compliance date of 31 December 2026, according to EY. It is not an AI rule, but it governs the data AI models depend on.

Sources: https://valor.globo.com/financas/noticia/2026/09/01/eventual-abalo-de-uma-prestadora-de-servicos-de-ia-relevante-poderia-colocar-estabilidade-financeira-em-risco-diz-bc.ghtml ; https://finsidersbrasil.com.br/tecnologia-para-fintechs/bc-estuda-regulacao-para-ia-e-criptografia-quantica/ ; https://ciberlatam.com/en/news/brazil-s-pl-2338-2023 ; https://www.ey.com/pt_br/newsroom/2026/08/nova-resolucao-bc-adocao-ia-exigem-gestao-eficiente-dados-instituicoes-financeiras

Mexico - a central bank warning, no federal AI law

Banco de Mexico published an AI and financial stability box in its June 2026 Financial Stability Report, dated 10 June 2026. It says AI does not always create wholly new risks, but it changes their scale, speed and interdependence, and it separates risks inherent in adoption from risks that come from the AI industry's own dependencies. That is a supervisory signal, not a rule. Mexico has no federal AI law. A cross-party Senate AI bill, dated 19 April 2026, is reported to be under committee review, but the source says the official information sheet was not yet available. In the Chamber, a 15 April 2026 item in the Gaceta Parlamentaria is a non-binding request to open working tables on a national AI law. The CNBV's president said on 25 September 2026, as reported by Forbes Mexico, that the regulator will seek more modern regulation for AI in the financial system and is using AI in its own supervision.

Sources: https://www.banxico.org.mx/publicaciones-y-prensa/reportes-sobre-el-sistema-financiero/recuadros/%7B564FE467-F838-EC6B-1615-39EB53E80B61%7D.pdf ; https://gaceta.diputados.gob.mx/Gaceta/66/2026/abr/20260415-III-4.html ; https://www.observatorio-ia-mexico.com/en/hemeroteca/primera-ley-de-inteligencia-artificial-del-senado-marco-integral-multipartidista ; https://forbes.com.mx/la-cnbv-se-compromete-a-una-regulacion-moderna-en-la-era-de-la-ia/

Chile - a framework bill that is about to change shape

Chile's AI bill (bulletin 16821-19, merged with 15869-19) was approved by the Chamber in October 2025 and is in the Senate's Future Challenges Committee. On 26 May 2026 the Science Minister told that committee the Executive would back a framework law built on six pillars, and the committee chair said it would probably arrive as a substitute amendment. On 2 October 2026 TVN reported that the Government will file a substitute amendment to replace the bill's risk classifications with a flexible framework law. Banks should not build to the old risk tiers as if they were final.

Sources: https://www.senado.cl/comunicaciones/noticias/senadores-conocen-propuesta-del-ejecutivo-para-ley-marco-de-ia ; https://www.tvn.cl/exponencial/noticias/ministra-lincolao-alista-cambios-al-proyecto-de-ia-queremos-habilitar ; https://iaexperta.cl/ia-legal/ley-inteligencia-artificial-chile/

Colombia - a national policy, supervisors using AI, and a new bill

Colombia has a national AI policy, CONPES 4144, and the planning department DNP leads its implementation. On 31 December 2025 DNP promoted an ethical AI guide for the public sector. That guide is for state entities, not banks. The Financial Superintendence has said it is using AI to supervise money laundering risk, in an October 2025 publication. On 21 July 2026 a group of Pacto Historico members filed a bill in the Chamber to regulate AI in Colombia. It is a filed bill, not law, and the sources checked show no AI-specific bank rule from the Superintendence.

Sources: https://www.dnp.gov.co/Prensa_/Noticias/Paginas/gobierno-impulsa-guia-etica-de-inteligencia-artificial.aspx ; https://www.superfinanciera.gov.co/publicaciones/10115843/superfinanciera-revoluciona-la-supervision-de-riesgos-de-lavado-de-activos-con-inteligencia-artificial/ ; https://www.camara.gov.co/wp-content/uploads/2026/07/proyectos-ley/documentos/proyecto-36127/P.L.025-2026SC-INTELIGENCIA-ARTIFICIAL.pdf

Ecuador - a data protection rule for AI systems

Ecuador's data protection superintendence issued Resolution SPDP-SPD-2026-0009-R, a general rule on protecting personal data in the use of AI systems. It builds on the data protection law, including the right not to be subject to decisions based solely or partly on automated assessments. For a bank or lender, the point is automated decisions about customers. The source is a legal news report of the resolution, not the resolution itself.

Source: https://www.lexis.com.ec/noticias/superintendencia-de-proteccion-de-datos-personales-expide-norma-general-para-la-proteccion-de-datos-personales-en-el-uso-de-inteligencia-artificial

Panama - a rejected framework bill

Reported, not regulator-confirmed: Nexo Capital reports that Panama's Education Committee rejected Bill 413, the country's first comprehensive AI framework bill, on 15 April 2026 after six months in subcommittee. The bill had four risk levels. If that report is right, Panama has no AI framework law, and any bank there is working from data protection and existing banking rules. This comes from press analysis and has not been checked against the Assembly record.

Source: https://nexocapital.pa/articulo/regulacion-ia-negado-comision-pl-413

Argentina and Uruguay - nothing binding found

The sources checked for this issue show no AI-specific rule for banks from the Central Bank of Argentina or the Central Bank of Uruguay. That is a scoped finding, not proof that nothing exists. Argentine and Uruguayan banks still answer to their general technology, outsourcing and data duties.

Why this matters beyond Latin America

The region shows three stages side by side. Peru has a law with a credit scoring rule. Brazil and Chile have bills that could change before they pass. Mexico and Colombia have central bank and policy signals. Compare it with Africa, where Kenya drew a borderline, and with Europe, where a binding law moved its date. The common question from supervisors is the same: who owns the model, and who cleared it?

What bank boards should do now

1. If you score credit customers in Peru, treat that use as high risk now. Check algorithmic transparency under Article 25 and the ability to explain a decision.

2. Do not wait for Brazil's bill. Build the AI inventory, model validation and vendor concentration record the Central Bank is already asking questions about.

3. Keep data lineage in scope. Joint Resolution 18's 31 December 2026 date covers the data your models use.

4. Hold Chile's risk tiers loosely until the substitute text is filed.

5. For Mexico, Colombia, Ecuador and Panama, record the supervisory signal and the pending bill, and review them each quarter.

Next issue: North America.

Source and status note

Positions checked against official and institutional materials available on 11 October 2026. Peru is from El Peruano and gob.pe; BDO Peru is the source for the 10 September 2026 deadline. Mexico's Banxico paper and the Colombian bill are from the institutions' own sites. Brazil's Central Bank remarks, the Mexican Senate bill, Chile's latest changes, Ecuador's resolution and Panama's bill rest on press or legal-news reports of official acts, and are marked as such. Panama is reported, not regulator-confirmed. This is general information, not legal advice.

THE AI REGULATION MARKET MAP • LATIN AMERICA • RAW COMPLIANCE

Seven markets. One binding AI rulebook for lenders, and it is not Brazil's.

Oonagh van den Berg | RAW Compliance

Issue 4 mapped Europe, where one AI Act met a long tail of national supervisors. Latin America has the opposite shape. Most of the region is still debating bills and publishing central bank warnings. One country has already written credit scoring into its AI rules.

That country is Peru. Brazil, the region's largest financial market, has a bill that has not reached a Chamber vote and a central bank that says it is studying the topic. A bank cannot treat "LATAM" as one jurisdiction. It has one market with live duties, several with pending bills, and a few where the honest answer is that nothing binding has been found.

Peru - the reglamento that names credit scoring

Peru's Regulation of Law 31814 was approved by Supreme Decree 115-2025-PCM and published in El Peruano on 9 September 2025. It takes a risk-based approach. Among the uses it treats as high risk is determining the credit assessment of persons, unless the AI is used to detect financial fraud. It also lists AI used to manage critical assets that support essential services, banking among them. Developers and deployers of high-risk systems must provide algorithmic transparency under Article 25, which includes explaining the criteria behind automated decisions that affect people's rights. The decree enters into force 90 business days after publication, apart from four final provisions that took effect the next day. BDO Peru reports that the first compliance deadline for private developers and deployers in six sectors, including economy and finance, fell on 10 September 2026, with later dates for other sectors. A lender using AI to score customers in Peru should read this as a live rule, not a consultation.

Sources: https://busquedas.elperuano.pe/dispositivo/NL/2436426-1 ; https://www.gob.pe/institucion/pcm/normas-legales/7133522-115-2025-pcm ; https://www.bdo.com.pe/es-pe/publicaciones/consultoria-de-negocios/2026/del-uso-de-ia-a-su-gobernanza-claves-para-llegar-preparados-al-10-de-setiembre

Brazil - a bill in the Chamber and a central bank that wants the sector role

PL 2338/2023, the AI legal framework, was approved by the Senate in December 2024 and sits in a special committee of the Chamber of Deputies. Trackers cited by CiberLATAM say the committee has no report and there is no official vote date. As drafted, it would create a national AI regulation and governance system coordinated by the data protection authority, ANPD, with the National Monetary Council and the Central Bank as sector regulators for finance. On 1 September 2026, a coordinator in the Central Bank's financial system regulation department said at a Brasilia event, as reported by Valor, that Brazil has no AI-specific regulation, that the Bank sees itself as the sector regulator if the framework passes, and that it is worried about concentration among AI service providers used by regulated firms. He also said the Bank is looking at how risk-modelling practices absorb AI risks. In March 2026 the same official told another event that the Bank was studying AI regulation. Separately, Joint Resolution 18 of the CMN and the Central Bank, in force from 1 January 2026, requires a data quality policy with traceability and named accountability, with a final compliance date of 31 December 2026, according to EY. It is not an AI rule, but it governs the data AI models depend on.

Sources: https://valor.globo.com/financas/noticia/2026/09/01/eventual-abalo-de-uma-prestadora-de-servicos-de-ia-relevante-poderia-colocar-estabilidade-financeira-em-risco-diz-bc.ghtml ; https://finsidersbrasil.com.br/tecnologia-para-fintechs/bc-estuda-regulacao-para-ia-e-criptografia-quantica/ ; https://ciberlatam.com/en/news/brazil-s-pl-2338-2023 ; https://www.ey.com/pt_br/newsroom/2026/08/nova-resolucao-bc-adocao-ia-exigem-gestao-eficiente-dados-instituicoes-financeiras

Mexico - a central bank warning, no federal AI law

Banco de Mexico published an AI and financial stability box in its June 2026 Financial Stability Report, dated 10 June 2026. It says AI does not always create wholly new risks, but it changes their scale, speed and interdependence, and it separates risks inherent in adoption from risks that come from the AI industry's own dependencies. That is a supervisory signal, not a rule. Mexico has no federal AI law. A cross-party Senate AI bill, dated 19 April 2026, is reported to be under committee review, but the source says the official information sheet was not yet available. In the Chamber, a 15 April 2026 item in the Gaceta Parlamentaria is a non-binding request to open working tables on a national AI law. The CNBV's president said on 25 September 2026, as reported by Forbes Mexico, that the regulator will seek more modern regulation for AI in the financial system and is using AI in its own supervision.

Sources: https://www.banxico.org.mx/publicaciones-y-prensa/reportes-sobre-el-sistema-financiero/recuadros/%7B564FE467-F838-EC6B-1615-39EB53E80B61%7D.pdf ; https://gaceta.diputados.gob.mx/Gaceta/66/2026/abr/20260415-III-4.html ; https://www.observatorio-ia-mexico.com/en/hemeroteca/primera-ley-de-inteligencia-artificial-del-senado-marco-integral-multipartidista ; https://forbes.com.mx/la-cnbv-se-compromete-a-una-regulacion-moderna-en-la-era-de-la-ia/

Chile - a framework bill that is about to change shape

Chile's AI bill (bulletin 16821-19, merged with 15869-19) was approved by the Chamber in October 2025 and is in the Senate's Future Challenges Committee. On 26 May 2026 the Science Minister told that committee the Executive would back a framework law built on six pillars, and the committee chair said it would probably arrive as a substitute amendment. On 2 October 2026 TVN reported that the Government will file a substitute amendment to replace the bill's risk classifications with a flexible framework law. Banks should not build to the old risk tiers as if they were final.

Sources: https://www.senado.cl/comunicaciones/noticias/senadores-conocen-propuesta-del-ejecutivo-para-ley-marco-de-ia ; https://www.tvn.cl/exponencial/noticias/ministra-lincolao-alista-cambios-al-proyecto-de-ia-queremos-habilitar ; https://iaexperta.cl/ia-legal/ley-inteligencia-artificial-chile/

Colombia - a national policy, supervisors using AI, and a new bill

Colombia has a national AI policy, CONPES 4144, and the planning department DNP leads its implementation. On 31 December 2025 DNP promoted an ethical AI guide for the public sector. That guide is for state entities, not banks. The Financial Superintendence has said it is using AI to supervise money laundering risk, in an October 2025 publication. On 21 July 2026 a group of Pacto Historico members filed a bill in the Chamber to regulate AI in Colombia. It is a filed bill, not law, and the sources checked show no AI-specific bank rule from the Superintendence.

Sources: https://www.dnp.gov.co/Prensa_/Noticias/Paginas/gobierno-impulsa-guia-etica-de-inteligencia-artificial.aspx ; https://www.superfinanciera.gov.co/publicaciones/10115843/superfinanciera-revoluciona-la-supervision-de-riesgos-de-lavado-de-activos-con-inteligencia-artificial/ ; https://www.camara.gov.co/wp-content/uploads/2026/07/proyectos-ley/documentos/proyecto-36127/P.L.025-2026SC-INTELIGENCIA-ARTIFICIAL.pdf

Ecuador - a data protection rule for AI systems

Ecuador's data protection superintendence issued Resolution SPDP-SPD-2026-0009-R, a general rule on protecting personal data in the use of AI systems. It builds on the data protection law, including the right not to be subject to decisions based solely or partly on automated assessments. For a bank or lender, the point is automated decisions about customers. The source is a legal news report of the resolution, not the resolution itself.

Source: https://www.lexis.com.ec/noticias/superintendencia-de-proteccion-de-datos-personales-expide-norma-general-para-la-proteccion-de-datos-personales-en-el-uso-de-inteligencia-artificial

Panama - a rejected framework bill

Reported, not regulator-confirmed: Nexo Capital reports that Panama's Education Committee rejected Bill 413, the country's first comprehensive AI framework bill, on 15 April 2026 after six months in subcommittee. The bill had four risk levels. If that report is right, Panama has no AI framework law, and any bank there is working from data protection and existing banking rules. This comes from press analysis and has not been checked against the Assembly record.

Source: https://nexocapital.pa/articulo/regulacion-ia-negado-comision-pl-413

Argentina and Uruguay - nothing binding found

The sources checked for this issue show no AI-specific rule for banks from the Central Bank of Argentina or the Central Bank of Uruguay. That is a scoped finding, not proof that nothing exists. Argentine and Uruguayan banks still answer to their general technology, outsourcing and data duties.

Why this matters beyond Latin America

The region shows three stages side by side. Peru has a law with a credit scoring rule. Brazil and Chile have bills that could change before they pass. Mexico and Colombia have central bank and policy signals. Compare it with Africa, where Kenya drew a borderline, and with Europe, where a binding law moved its date. The common question from supervisors is the same: who owns the model, and who cleared it?

What bank boards should do now

1. If you score credit customers in Peru, treat that use as high risk now. Check algorithmic transparency under Article 25 and the ability to explain a decision.

2. Do not wait for Brazil's bill. Build the AI inventory, model validation and vendor concentration record the Central Bank is already asking questions about.

3. Keep data lineage in scope. Joint Resolution 18's 31 December 2026 date covers the data your models use.

4. Hold Chile's risk tiers loosely until the substitute text is filed.

5. For Mexico, Colombia, Ecuador and Panama, record the supervisory signal and the pending bill, and review them each quarter.

Next issue: North America.

Source and status note

Positions checked against official and institutional materials available on 11 October 2026. Peru is from El Peruano and gob.pe; BDO Peru is the source for the 10 September 2026 deadline. Mexico's Banxico paper and the Colombian bill are from the institutions' own sites. Brazil's Central Bank remarks, the Mexican Senate bill, Chile's latest changes, Ecuador's resolution and Panama's bill rest on press or legal-news reports of official acts, and are marked as such. Panama is reported, not regulator-confirmed. This is general information, not legal advice.

THE AI REGULATION MARKET MAP • LATIN AMERICA • RAW COMPLIANCE

Seven markets. One binding AI rulebook for lenders, and it is not Brazil's.

Oonagh van den Berg | RAW Compliance

Issue 4 mapped Europe, where one AI Act met a long tail of national supervisors. Latin America has the opposite shape. Most of the region is still debating bills and publishing central bank warnings. One country has already written credit scoring into its AI rules.

That country is Peru. Brazil, the region's largest financial market, has a bill that has not reached a Chamber vote and a central bank that says it is studying the topic. A bank cannot treat "LATAM" as one jurisdiction. It has one market with live duties, several with pending bills, and a few where the honest answer is that nothing binding has been found.

Peru - the reglamento that names credit scoring

Peru's Regulation of Law 31814 was approved by Supreme Decree 115-2025-PCM and published in El Peruano on 9 September 2025. It takes a risk-based approach. Among the uses it treats as high risk is determining the credit assessment of persons, unless the AI is used to detect financial fraud. It also lists AI used to manage critical assets that support essential services, banking among them. Developers and deployers of high-risk systems must provide algorithmic transparency under Article 25, which includes explaining the criteria behind automated decisions that affect people's rights. The decree enters into force 90 business days after publication, apart from four final provisions that took effect the next day. BDO Peru reports that the first compliance deadline for private developers and deployers in six sectors, including economy and finance, fell on 10 September 2026, with later dates for other sectors. A lender using AI to score customers in Peru should read this as a live rule, not a consultation.

Sources: https://busquedas.elperuano.pe/dispositivo/NL/2436426-1 ; https://www.gob.pe/institucion/pcm/normas-legales/7133522-115-2025-pcm ; https://www.bdo.com.pe/es-pe/publicaciones/consultoria-de-negocios/2026/del-uso-de-ia-a-su-gobernanza-claves-para-llegar-preparados-al-10-de-setiembre

Brazil - a bill in the Chamber and a central bank that wants the sector role

PL 2338/2023, the AI legal framework, was approved by the Senate in December 2024 and sits in a special committee of the Chamber of Deputies. Trackers cited by CiberLATAM say the committee has no report and there is no official vote date. As drafted, it would create a national AI regulation and governance system coordinated by the data protection authority, ANPD, with the National Monetary Council and the Central Bank as sector regulators for finance. On 1 September 2026, a coordinator in the Central Bank's financial system regulation department said at a Brasilia event, as reported by Valor, that Brazil has no AI-specific regulation, that the Bank sees itself as the sector regulator if the framework passes, and that it is worried about concentration among AI service providers used by regulated firms. He also said the Bank is looking at how risk-modelling practices absorb AI risks. In March 2026 the same official told another event that the Bank was studying AI regulation. Separately, Joint Resolution 18 of the CMN and the Central Bank, in force from 1 January 2026, requires a data quality policy with traceability and named accountability, with a final compliance date of 31 December 2026, according to EY. It is not an AI rule, but it governs the data AI models depend on.

Sources: https://valor.globo.com/financas/noticia/2026/09/01/eventual-abalo-de-uma-prestadora-de-servicos-de-ia-relevante-poderia-colocar-estabilidade-financeira-em-risco-diz-bc.ghtml ; https://finsidersbrasil.com.br/tecnologia-para-fintechs/bc-estuda-regulacao-para-ia-e-criptografia-quantica/ ; https://ciberlatam.com/en/news/brazil-s-pl-2338-2023 ; https://www.ey.com/pt_br/newsroom/2026/08/nova-resolucao-bc-adocao-ia-exigem-gestao-eficiente-dados-instituicoes-financeiras

Mexico - a central bank warning, no federal AI law

Banco de Mexico published an AI and financial stability box in its June 2026 Financial Stability Report, dated 10 June 2026. It says AI does not always create wholly new risks, but it changes their scale, speed and interdependence, and it separates risks inherent in adoption from risks that come from the AI industry's own dependencies. That is a supervisory signal, not a rule. Mexico has no federal AI law. A cross-party Senate AI bill, dated 19 April 2026, is reported to be under committee review, but the source says the official information sheet was not yet available. In the Chamber, a 15 April 2026 item in the Gaceta Parlamentaria is a non-binding request to open working tables on a national AI law. The CNBV's president said on 25 September 2026, as reported by Forbes Mexico, that the regulator will seek more modern regulation for AI in the financial system and is using AI in its own supervision.

Sources: https://www.banxico.org.mx/publicaciones-y-prensa/reportes-sobre-el-sistema-financiero/recuadros/%7B564FE467-F838-EC6B-1615-39EB53E80B61%7D.pdf ; https://gaceta.diputados.gob.mx/Gaceta/66/2026/abr/20260415-III-4.html ; https://www.observatorio-ia-mexico.com/en/hemeroteca/primera-ley-de-inteligencia-artificial-del-senado-marco-integral-multipartidista ; https://forbes.com.mx/la-cnbv-se-compromete-a-una-regulacion-moderna-en-la-era-de-la-ia/

Chile - a framework bill that is about to change shape

Chile's AI bill (bulletin 16821-19, merged with 15869-19) was approved by the Chamber in October 2025 and is in the Senate's Future Challenges Committee. On 26 May 2026 the Science Minister told that committee the Executive would back a framework law built on six pillars, and the committee chair said it would probably arrive as a substitute amendment. On 2 October 2026 TVN reported that the Government will file a substitute amendment to replace the bill's risk classifications with a flexible framework law. Banks should not build to the old risk tiers as if they were final.

Sources: https://www.senado.cl/comunicaciones/noticias/senadores-conocen-propuesta-del-ejecutivo-para-ley-marco-de-ia ; https://www.tvn.cl/exponencial/noticias/ministra-lincolao-alista-cambios-al-proyecto-de-ia-queremos-habilitar ; https://iaexperta.cl/ia-legal/ley-inteligencia-artificial-chile/

Colombia - a national policy, supervisors using AI, and a new bill

Colombia has a national AI policy, CONPES 4144, and the planning department DNP leads its implementation. On 31 December 2025 DNP promoted an ethical AI guide for the public sector. That guide is for state entities, not banks. The Financial Superintendence has said it is using AI to supervise money laundering risk, in an October 2025 publication. On 21 July 2026 a group of Pacto Historico members filed a bill in the Chamber to regulate AI in Colombia. It is a filed bill, not law, and the sources checked show no AI-specific bank rule from the Superintendence.

Sources: https://www.dnp.gov.co/Prensa_/Noticias/Paginas/gobierno-impulsa-guia-etica-de-inteligencia-artificial.aspx ; https://www.superfinanciera.gov.co/publicaciones/10115843/superfinanciera-revoluciona-la-supervision-de-riesgos-de-lavado-de-activos-con-inteligencia-artificial/ ; https://www.camara.gov.co/wp-content/uploads/2026/07/proyectos-ley/documentos/proyecto-36127/P.L.025-2026SC-INTELIGENCIA-ARTIFICIAL.pdf

Ecuador - a data protection rule for AI systems

Ecuador's data protection superintendence issued Resolution SPDP-SPD-2026-0009-R, a general rule on protecting personal data in the use of AI systems. It builds on the data protection law, including the right not to be subject to decisions based solely or partly on automated assessments. For a bank or lender, the point is automated decisions about customers. The source is a legal news report of the resolution, not the resolution itself.

Source: https://www.lexis.com.ec/noticias/superintendencia-de-proteccion-de-datos-personales-expide-norma-general-para-la-proteccion-de-datos-personales-en-el-uso-de-inteligencia-artificial

Panama - a rejected framework bill

Reported, not regulator-confirmed: Nexo Capital reports that Panama's Education Committee rejected Bill 413, the country's first comprehensive AI framework bill, on 15 April 2026 after six months in subcommittee. The bill had four risk levels. If that report is right, Panama has no AI framework law, and any bank there is working from data protection and existing banking rules. This comes from press analysis and has not been checked against the Assembly record.

Source: https://nexocapital.pa/articulo/regulacion-ia-negado-comision-pl-413

Argentina and Uruguay - nothing binding found

The sources checked for this issue show no AI-specific rule for banks from the Central Bank of Argentina or the Central Bank of Uruguay. That is a scoped finding, not proof that nothing exists. Argentine and Uruguayan banks still answer to their general technology, outsourcing and data duties.

Why this matters beyond Latin America

The region shows three stages side by side. Peru has a law with a credit scoring rule. Brazil and Chile have bills that could change before they pass. Mexico and Colombia have central bank and policy signals. Compare it with Africa, where Kenya drew a borderline, and with Europe, where a binding law moved its date. The common question from supervisors is the same: who owns the model, and who cleared it?

What bank boards should do now

1. If you score credit customers in Peru, treat that use as high risk now. Check algorithmic transparency under Article 25 and the ability to explain a decision.

2. Do not wait for Brazil's bill. Build the AI inventory, model validation and vendor concentration record the Central Bank is already asking questions about.

3. Keep data lineage in scope. Joint Resolution 18's 31 December 2026 date covers the data your models use.

4. Hold Chile's risk tiers loosely until the substitute text is filed.

5. For Mexico, Colombia, Ecuador and Panama, record the supervisory signal and the pending bill, and review them each quarter.

Next issue: North America.

Source and status note

Positions checked against official and institutional materials available on 11 October 2026. Peru is from El Peruano and gob.pe; BDO Peru is the source for the 10 September 2026 deadline. Mexico's Banxico paper and the Colombian bill are from the institutions' own sites. Brazil's Central Bank remarks, the Mexican Senate bill, Chile's latest changes, Ecuador's resolution and Panama's bill rest on press or legal-news reports of official acts, and are marked as such. Panama is reported, not regulator-confirmed. This is general information, not legal advice.

THE AI REGULATION MARKET MAP • LATIN AMERICA • RAW COMPLIANCE