https://sled.lighthouseadvisory.consulting/feeds/aws.xml2026-11-11T05:15:55.646ZLighthouse Advisoryhttps://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-102026-11-11T05:15:55.646Z2026-11-10T00:00:00.000Z

November 10 AWS news centered on what SLED buyers must decide for themselves. AWS promoted Moonshot AI's Kimi K3.1 Pro on Bedrock, saying data stays within AWS and is not shared with Moonshot, while Texas has listed Moonshot AI as a prohibited technology for state employees and devices since January, without saying how that applies to cloud-hosted models. OpenAI's Daybreak cyber models reached Bedrock's runtime endpoint for enrolled customers; an independent Irregular evaluation of the model behind Daybreak Blue found real zero-day discovery but low end-to-end reliability. AWS also released an AgentCore Web Search API (commercial Regions only, not GovCloud), an agent evaluation playbook, and a token-cost data lake that centralizes full prompts that may contain PII. In Ohio, Butler County planners backed a township ban that could block a proposed $5 billion AWS campus.

November 10 AWS news centered on what SLED buyers must decide for themselves. AWS promoted Moonshot AI's Kimi K3.1 Pro on Bedrock, saying data stays within AWS and is not shared with Moonshot, while Texas has listed Moonshot AI as a prohibited technology for state employees and devices since January, without saying how that applies to cloud-hosted models. OpenAI's Daybreak cyber models reached Bedrock's runtime endpoint for enrolled customers; an independent Irregular evaluation of the model behind Daybreak Blue found real zero-day discovery but low end-to-end reliability. AWS also released an AgentCore Web Search API (commercial Regions only, not GovCloud), an agent evaluation playbook, and a token-cost data lake that centralizes full prompts that may contain PII. In Ohio, Butler County planners backed a township ban that could block a proposed $5 billion AWS campus. 8 sources · 2 cross-source patterns Bedrock's expanding model catalog moves vendor-eligibility and cyber-access decisions onto SLED buyers Who reviews each new Bedrock model against state vendor restrictions, third-party access programs and independent evidence before it is allowed in agency accounts? New AgentCore and Bedrock operations tooling arrives commercial-first, and evaluation thresholds, prompt archives and query budgets stay with the customer For each production agent, who sets evaluation thresholds, approves grounding domains and query budgets, and decides how long prompt logs are kept, and is the tooling available in the partition the agent runs in?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-092026-11-10T07:05:16.753Z2026-11-09T00:00:00.000Z

November 9 brought new Anthropic models to AWS GovCloud (US), each with conditions attached. Claude Fable 5.2 is available there as a 'covered model' that requires data retention for safety monitoring unless the customer qualifies for zero retention. Claude Mythos 5.2, a cybersecurity model, is limited to organizations allow-listed through Anthropic's Cyber Verification Program, whose Bedrock access currently depends on a data-retention exemption. AWS's FedRAMP/IL model status page, also updated that day, did not list either model, and independent reporting notes no benchmarks or eligibility criteria were published. AWS also shipped cross-partition federation between GovCloud and commercial accounts, advising customers to treat it as an exception, and published a sample memory-poisoning defense for AgentCore agents. In higher-education research, an AWS story profiles Northwestern-born de-identification and citation tools, but its savings figures are unverified.

November 9 brought new Anthropic models to AWS GovCloud (US), each with conditions attached. Claude Fable 5.2 is available there as a 'covered model' that requires data retention for safety monitoring unless the customer qualifies for zero retention. Claude Mythos 5.2, a cybersecurity model, is limited to organizations allow-listed through Anthropic's Cyber Verification Program, whose Bedrock access currently depends on a data-retention exemption. AWS's FedRAMP/IL model status page, also updated that day, did not list either model, and independent reporting notes no benchmarks or eligibility criteria were published. AWS also shipped cross-partition federation between GovCloud and commercial accounts, advising customers to treat it as an exception, and published a sample memory-poisoning defense for AgentCore agents. In higher-education research, an AWS story profiles Northwestern-born de-identification and citation tools, but its savings figures are unverified. 8 sources · 2 cross-source patterns New frontier models reach GovCloud with gates attached: verification, data retention and authorization listing Before any team uses a newly announced GovCloud model, who confirms its retention mode, its authorization listing at the required level and any third-party eligibility approval? AWS's November 9 security guidance leaves agent-memory and cross-partition controls for customers to assemble and approve Which team approves each agent memory namespace and each GovCloud-to-commercial trust, and how is that approval tested and reviewed?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-072026-11-08T05:04:18.486Z2026-11-07T00:00:00.000Z

No new measured AWS AI deployment results appeared overnight. The signals were workforce, siting and operational assurance. CGI Federal and AWS announced an AI workforce coalition with the University of Tennessee, Knoxville as first campus; independent reporting notes it has no timeline, enrollment or placement metrics yet. In Ohio, Amazon's Wilmington site plan goes to a November 10 planning vote with a staff approval recommendation, after a court-ordered pause and with permits and a 2027 statewide-ban push still pending. Fayette County's new-project pause exempts the AWS campus under construction. On the platform side, Security Hub added a no-added-cost SBOM inventory (available wherever Security Hub runs, including GovCloud) that can surface outdated AI SDK packages. A competing vendor's AgentCore review lists controls customers still own, and independent monitoring logged about 2.4 hours of Bedrock degradation tied to a US-West-2 EC2/EBS incident on November 6.

No new measured AWS AI deployment results appeared overnight. The signals were workforce, siting and operational assurance. CGI Federal and AWS announced an AI workforce coalition with the University of Tennessee, Knoxville as first campus; independent reporting notes it has no timeline, enrollment or placement metrics yet. In Ohio, Amazon's Wilmington site plan goes to a November 10 planning vote with a staff approval recommendation, after a court-ordered pause and with permits and a 2027 statewide-ban push still pending. Fayette County's new-project pause exempts the AWS campus under construction. On the platform side, Security Hub added a no-added-cost SBOM inventory (available wherever Security Hub runs, including GovCloud) that can surface outdated AI SDK packages. A competing vendor's AgentCore review lists controls customers still own, and independent monitoring logged about 2.4 hours of Bedrock degradation tied to a US-West-2 EC2/EBS incident on November 6. 7 sources · 2 cross-source patterns After the Nov 3 bans, AWS Ohio campuses now hinge on site-by-site municipal and county gates For each Ohio AWS campus your stakeholders reference, is it under construction, awaiting a local vote, or exposed to a future statewide ban, and which official record confirms that? AWS AI assurance work stays on the customer side: package inventory, identity binding and regional resilience Who in the agency owns AI SDK version policy, AgentCore session binding and Bedrock fallback behavior, and how is each tested before production?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-062026-11-07T05:48:20.635Z2026-11-06T00:00:00.000Z

AWS expanded AgentCore governance into GovCloud while Ohio municipalities kept tightening campus siting. Agent Registry became generally available in AWS GovCloud (US-West) as a curated catalog for MCP servers, tools, agents and skills with approval workflows and CloudTrail, and Briefpixel independently cautioned that a registry does not create automatic compliance or confirm US-East coverage. Journal-News and Fortune detail post-Nov 3 Ohio outcomes: Trenton's ~80% ban on >25 MW centers that organizers aim at a proposed 18-building AWS campus (while a four-building AWS site under construction likely continues), plus at least 14 Ohio municipalities passing bans or restrictions, including a Sunbury measure aimed at a possible Amazon project. AWS Bulletin 2026-065 remains relevant why-now: catalog expansion does not replace bedrock-agentcore >=1.18.1 and install_packages() input hygiene.

AWS expanded AgentCore governance into GovCloud while Ohio municipalities kept tightening campus siting. Agent Registry became generally available in AWS GovCloud (US-West) as a curated catalog for MCP servers, tools, agents and skills with approval workflows and CloudTrail, and Briefpixel independently cautioned that a registry does not create automatic compliance or confirm US-East coverage. Journal-News and Fortune detail post-Nov 3 Ohio outcomes: Trenton's ~80% ban on >25 MW centers that organizers aim at a proposed 18-building AWS campus (while a four-building AWS site under construction likely continues), plus at least 14 Ohio municipalities passing bans or restrictions, including a Sunbury measure aimed at a possible Amazon project. AWS Bulletin 2026-065 remains relevant why-now: catalog expansion does not replace bedrock-agentcore >=1.18.1 and install_packages() input hygiene. 5 sources · 2 cross-source patterns GovCloud Agent Registry expands governed discovery while AgentCore SDK bulletin still requires runtime hardening When standing up Agent Registry in GovCloud, how will curators prove registered agents run patched SDKs and least-privilege Code Interpreters? Ohio municipal bans tighten AWS campus siting after Nov 3, while under-construction sites and cloud Regions remain distinct Which customer or capacity commitments depend on contested Ohio parcels versus AWS Regions that are unaffected by municipal bans?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-052026-11-06T11:55:52.192Z2026-11-05T00:00:00.000Z

Fresh AWS product signals landed alongside a sharper post-election siting map. AWS made Web Answer generally available on Bedrock AgentCore as a managed, cited web-answer tool, but only in US East (N. Virginia) with no GovCloud claim, and brought Security Hub Network Scanning to GovCloud to probe actual internet reachability at no extra Essentials cost. Independently, KUOW and a Dealroom midterm scorecard describe Ohio local bans (Trenton ~80% against >25 MW centers aimed at an Amazon campus) and Texas/Georgia/Ohio gubernatorial pledges to pause or tighten data centers, with analysts warning short-term moratoria could touch a large share of planned US capacity—politics that agencies should not confuse with Bedrock or GovCloud service availability.

Fresh AWS product signals landed alongside a sharper post-election siting map. AWS made Web Answer generally available on Bedrock AgentCore as a managed, cited web-answer tool, but only in US East (N. Virginia) with no GovCloud claim, and brought Security Hub Network Scanning to GovCloud to probe actual internet reachability at no extra Essentials cost. Independently, KUOW and a Dealroom midterm scorecard describe Ohio local bans (Trenton ~80% against >25 MW centers aimed at an Amazon campus) and Texas/Georgia/Ohio gubernatorial pledges to pause or tighten data centers, with analysts warning short-term moratoria could touch a large share of planned US capacity—politics that agencies should not confuse with Bedrock or GovCloud service availability. 4 sources · 1 cross-source patterns Post-midterm politics escalate AWS data-center siting from local bans to statewide moratorium pledges, while AWS keeps shipping agent and GovCloud security features How will a jurisdiction separate cloud-service decisions (Bedrock/GovCloud) from hyperscale campus permitting under new post-election moratorium pledges?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-042026-11-05T05:44:07.516Z2026-11-04T00:00:00.000Z

Election night turned local resistance to AWS data centers into binding decisions. Sunbury, Ohio voters passed a 25 MW cap aimed at an Amazon campus 2,809 to 763, and Trenton's ban passed about 80-20, among at least 14 Ohio towns approving limits (unofficial results). In West Richland, Wash., the council adopted interim zoning limits 6-0 but a moratorium motion failed 3-3, and GeekWire reports only 23 of 706 comments on Amazon's Wallula Gap site were supportive as AWS's CEO warned moratoriums could cost the US the AI race. On the product side, AWS added Linux (with root access) and Windows 11 desktops to WorkSpaces for AI agents, without stating GovCloud availability.

Election night turned local resistance to AWS data centers into binding decisions. Sunbury, Ohio voters passed a 25 MW cap aimed at an Amazon campus 2,809 to 763, and Trenton's ban passed about 80-20, among at least 14 Ohio towns approving limits (unofficial results). In West Richland, Wash., the council adopted interim zoning limits 6-0 but a moratorium motion failed 3-3, and GeekWire reports only 23 of 706 comments on Amazon's Wallula Gap site were supportive as AWS's CEO warned moratoriums could cost the US the AI race. On the product side, AWS added Linux (with root access) and Windows 11 desktops to WorkSpaces for AI agents, without stating GovCloud availability. 4 sources · 1 cross-source patterns Local governments are moving from debate to binding decisions on AWS data centers, while AWS argues moratoriums threaten the AI race What written siting criteria and independent verification will a jurisdiction require before an AWS (or any hyperscale) proposal reaches a vote?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-032026-11-04T15:02:19.555Z2026-11-03T00:00:00.000Z

A thin Tuesday edition dominated by AWS's own material. AWS pitched AI to state tax and motor vehicle agencies, citing Kentucky, Alabama and Wisconsin contact-center gains it does not substantiate; added line-level requirements traceability to AWS Transform for mainframe for COBOL and PL/I modernization; and offered education leaders a framework for reviewing AI spend as introductory pricing ends. Independently, the Tri-Cities Area Journal of Business reports Amazon has filed no permits and begun no state environmental review for its West Richland, Wash., campus as the council weighed a six-month moratorium on Nov 3. Election-night outcomes in Sunbury, Ohio were not yet available.

A thin Tuesday edition dominated by AWS's own material. AWS pitched AI to state tax and motor vehicle agencies, citing Kentucky, Alabama and Wisconsin contact-center gains it does not substantiate; added line-level requirements traceability to AWS Transform for mainframe for COBOL and PL/I modernization; and offered education leaders a framework for reviewing AI spend as introductory pricing ends. Independently, the Tri-Cities Area Journal of Business reports Amazon has filed no permits and begun no state environmental review for its West Richland, Wash., campus as the council weighed a six-month moratorium on Nov 3. Election-night outcomes in Sunbury, Ohio were not yet available. 4 sources · 1 cross-source patterns AWS is marketing AI for core state legacy operations, but the evidence offered is vendor-reported speed and traceability claims What baseline, ground-truth test and independent review will the agency require before accepting vendor-reported AI modernization results?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-022026-11-03T11:43:19.728Z2026-11-02T00:00:00.000Z

A quiet Monday edition of product notices and local siting fights. AWS added Claude Haiku 5.5 to Kiro in GovCloud, giving developer teams there a cheaper model tier for bulk and review tasks; let Amazon Connect flows call MCP tools through AgentCore Gateway, but only in commercial Regions; and simplified Amazon Quick desktop sign-in. None reports a SLED customer or measured result. On the constraint side, the Tri-City Herald reports AWS confirmed it is behind a $264 million West Richland, Wash., land deal for up to 21 data center buildings as the council weighs a moratorium, and Sunbury, Ohio votes Nov 3 on a charter ban that could stop a proposed $2 billion Amazon campus. Company water and power claims in both remain unverified.

A quiet Monday edition of product notices and local siting fights. AWS added Claude Haiku 5.5 to Kiro in GovCloud, giving developer teams there a cheaper model tier for bulk and review tasks; let Amazon Connect flows call MCP tools through AgentCore Gateway, but only in commercial Regions; and simplified Amazon Quick desktop sign-in. None reports a SLED customer or measured result. On the constraint side, the Tri-City Herald reports AWS confirmed it is behind a $264 million West Richland, Wash., land deal for up to 21 data center buildings as the council weighs a moratorium, and Sunbury, Ohio votes Nov 3 on a charter ban that could stop a proposed $2 billion Amazon campus. Company water and power claims in both remain unverified. 5 sources · 1 cross-source patterns AWS data center expansion is meeting local moratoriums and ballot bans, with company utility claims left for councils to verify What independent review, disclosure and enforceable commitments does the local government require before approving a hyperscale campus?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-11-012026-11-03T03:22:04.815Z2026-11-01T00:00:00.000Z

A thin weekend window with one SLED story and one infrastructure story, both from AWS and both self-measured. AWS says rebuilding NASWA's National Labor Exchange job-posting extraction around Amazon Nova Lite and per-field prompts cut processing costs 80 percent, with parsed pay, benefits and education fields reaching state workforce users in 2027. AWS gives no baseline or accuracy figures; the only published validation is a peer-reviewed 2025 paper by Amazon and NASWA authors that checked one field against Census data. Separately, AWS reports its European Sovereign Cloud ran for several hours with its global backbone and operational-data transfer system switched off. An independent commentary notes the test was provider-run, short, and silent on customer continuity and jurisdiction. U.S. SLED relevance is indirect.

A thin weekend window with one SLED story and one infrastructure story, both from AWS and both self-measured. AWS says rebuilding NASWA's National Labor Exchange job-posting extraction around Amazon Nova Lite and per-field prompts cut processing costs 80 percent, with parsed pay, benefits and education fields reaching state workforce users in 2027. AWS gives no baseline or accuracy figures; the only published validation is a peer-reviewed 2025 paper by Amazon and NASWA authors that checked one field against Census data. Separately, AWS reports its European Sovereign Cloud ran for several hours with its global backbone and operational-data transfer system switched off. An independent commentary notes the test was provider-run, short, and silent on customer continuity and jurisdiction. U.S. SLED relevance is indirect. 4 sources · 1 cross-source patterns AWS's in-window results are self-measured, and the outside checks available are narrower than the claims For each AWS-reported result you might rely on, what baseline, method and independent check exist, and what would you need to measure yourself?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-10-312026-11-02T01:21:42.732Z2026-10-31T00:00:00.000Z

A GovCloud-and-guardrails edition. On Oct 30 AWS documented how to run the ChatGPT desktop app and Codex against Bedrock in GovCloud US-West under a locked-down configuration, made OpenAI's Daybreak Red cyber model available to vetted customers in GovCloud US-East, and brought AWS Transform migration and second-generation Outposts racks into GovCloud. Each launch leaves authorization, eligibility or feature gaps with the customer, and none reports a SLED deployment. On the constraint side, an AWS bulletin on the deprecated AgentCore Starter Toolkit requires regenerating affected agents, and an independent preprint shows prompt-caching mistakes, including Bedrock-specific ones, mostly fail silently. Virginia Tech's Amazon-funded ML initiative funds agent reliability research, and WVXU reports the water, power and zoning questions facing a proposed AWS campus ahead of Trenton, Ohio's Nov 3 data center ban vote.

A GovCloud-and-guardrails edition. On Oct 30 AWS documented how to run the ChatGPT desktop app and Codex against Bedrock in GovCloud US-West under a locked-down configuration, made OpenAI's Daybreak Red cyber model available to vetted customers in GovCloud US-East, and brought AWS Transform migration and second-generation Outposts racks into GovCloud. Each launch leaves authorization, eligibility or feature gaps with the customer, and none reports a SLED deployment. On the constraint side, an AWS bulletin on the deprecated AgentCore Starter Toolkit requires regenerating affected agents, and an independent preprint shows prompt-caching mistakes, including Bedrock-specific ones, mostly fail silently. Virginia Tech's Amazon-funded ML initiative funds agent reliability research, and WVXU reports the water, power and zoning questions facing a proposed AWS campus ahead of Trenton, Ohio's Nov 3 data center ban vote. 7 sources · 2 cross-source patterns AWS is moving frontier models, agents and migration automation into GovCloud, but every launch leaves authorization, eligibility or feature gaps with the customer For each new GovCloud AI capability, which Region, features and approvals does the agency actually get, and what still has to be authorized or done outside the boundary? AWS agent tooling can go wrong without visible errors, so teams need inventory and re-checks after tool or provider changes Do you know which tooling versions and provider settings built each production agent, and do you re-test after AWS changes them?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-10-302026-10-31T04:40:43.122Z2026-10-30T00:00:00.000Z

AWS SLED edition, October 30, 2026. Agent security dominates. Independent Zenity research found that, until late September, one prompt to an exposed Bedrock AgentCore agent could expose credentials, conversations and secrets across an account-region. AWS disputes the framing. Meanwhile, AWS's new guidance says model-only boundaries are not controls, and its CJIS RAG design for municipal police stresses that retrieval filtering is not authorization. GovCloud gained OpenAI GPT-5.6 with cross-Region inference. Kansas City-area counties are using Granicus's Bedrock-based assistant, with modest vendor-reported results and transparency gaps. Virginia Tech and Pitt expanded AWS-backed AI education programs. Five of seven items are vendor or institutional material.

AWS SLED edition, October 30, 2026. Agent security dominates. Independent Zenity research found that, until late September, one prompt to an exposed Bedrock AgentCore agent could expose credentials, conversations and secrets across an account-region. AWS disputes the framing. Meanwhile, AWS's new guidance says model-only boundaries are not controls, and its CJIS RAG design for municipal police stresses that retrieval filtering is not authorization. GovCloud gained OpenAI GPT-5.6 with cross-Region inference. Kansas City-area counties are using Granicus's Bedrock-based assistant, with modest vendor-reported results and transparency gaps. Virginia Tech and Pitt expanded AWS-backed AI education programs. Five of seven items are vendor or institutional material. 7 sources · 2 cross-source patterns Agent and retrieval boundaries must be enforced outside the model, and vendor defaults verified For each agent or RAG pilot, which limits are enforced by IAM, policy or application identity rather than by the model? Have we checked execution roles against current, not assumed, defaults? AWS-backed campus programs are framed as AI workforce pipelines, without outcome measures If we join or replicate a vendor-supported AI education program, what learning or adoption outcomes will we measure, and who sustains the outputs?https://sled.lighthouseadvisory.consulting/streams/aws/editions/2026-10-292026-10-30T03:22:01.489Z2026-10-29T00:00:00.000Z

AWS SLED edition, October 29, 2026. GovCloud model options widened with Claude Haiku 5.5 and Grok 4.7, but no public-sector evaluations accompany them. Texas presented its SAM regulatory assistant as an AI service in AWS GovCloud, while Richmond's Amazon Connect nonemergency line handles about 19% of calls without a dispatcher against a 50-60% goal. AWS guidance for HHS eligibility, EdTech and higher-ed assistants consistently keeps humans on final decisions and puts guardrails and evaluation before scale. Evidence is thin: seven of nine items are vendor material.

AWS SLED edition, October 29, 2026. GovCloud model options widened with Claude Haiku 5.5 and Grok 4.7, but no public-sector evaluations accompany them. Texas presented its SAM regulatory assistant as an AI service in AWS GovCloud, while Richmond's Amazon Connect nonemergency line handles about 19% of calls without a dispatcher against a 50-60% goal. AWS guidance for HHS eligibility, EdTech and higher-ed assistants consistently keeps humans on final decisions and puts guardrails and evaluation before scale. Evidence is thin: seven of nine items are vendor material. 9 sources · 3 cross-source patterns GovCloud model choice is widening faster than agencies can evaluate it Do we have a standing, repeatable process to evaluate and approve each new GovCloud model on our own tasks before it reaches production? Public-facing assistants: designs stress guardrails, while the only measured outcome falls short What containment, accuracy and disclosure metrics will we commit to measuring from day one of a public-facing assistant? Humans keep final decisions in AWS SLED guidance For each AI workflow, which decisions remain human, and what measured evidence would justify changing that boundary?