AWS Builder Center

AWS re:Invent 2026: The Attendee Playbook - Security Track: AI Security

...

Part of the AWS re:Invent 2026 attendee blog series - start at the  index  for venue map, transit times, Expo strategy and the persona matrix.
Who it's for: security engineers and architects securing AI agents, AI and platform teams shipping agents to production, CISOs and AI-risk owners, and GRC leads facing the EU AI Act. AI changes security twice: attackers move at machine speed, and your own agents become privileged actors that need identities, permissions, guardrails and an audit trail. Most AI-security sessions run at Wynn/Encore, with agent-runtime and guardrail deep dives at MGM Grand, so the plan stays North and goes South as a labeled alternative. For IAM, threat detection, encryption and compliance, see the  security & compliance post . All sessions come from the official catalog (read Nov 12, 2026).
Career tracks
  • Early career: SEC328-R (AI guardrails across the organization), SEC302 and SEC365.
  • Mid career: SEC362 (agent identity), SEC382 (prompt injection), SEC336, SEC327 and SEC321-R (AI coding agents).
  • Senior / AI-risk owners: SEC407-R (level 400), SEC409, INV525 (level 500), SEC386 and the SEC003 mainstage on security in a world of AI.
Must-attend keynotes: CEO Keynote – Matt Garman (8:30 AM, The Venetian) for agent and security announcements. On Thursday, skip the Technology Innovation Keynote for the Wynn AI-security path and watch the replay.

What to see

New launches

  • What's new: SEC398 (the latest innovations in multicloud agent security).
  • AWS Continuum (the 2026 home of AWS Security Agent): the SEC312 workshop moves you from annual pentests to continuous testing. More Continuum sessions are in the  security & compliance post .
Keynotes will add sessions, so check the catalog's new-launch filter after each keynote and search for "[NEW LAUNCH]".

Established products

Grouped by the four questions to answer before an agent reaches production.
  • 1. Securing AI agents. Permissions and Cedar: the SEC407-R workshop (repeats SEC407-R1 at Caesars Forum) builds scoped execution roles with permissions boundaries, enforces Amazon Bedrock Guardrails through IAM policy, validates with IAM Access Analyzer and adds Cedar enforcement in AgentCore. Identity: SEC362, SEC344, SEC419 (AgentCore Identity) and PEX401-R1. Multi-agent and MCP: SEC327, SEC336, SEC343, NET321-R1 and SEC352-R (repeats SEC352-R1). Isolation and detection: INV525, SEC429-R and SEC409. Security agents: SEC310 (lessons from AWS at scale).
  • 2. Guardrails and model safety. SEC328-R, AIM453 (Bedrock Guardrails for the agent loop), AIM451 (automated reasoning), SEC382, the SEC341 workshop, the SEC403 red-team CTF, INV503-R1 and SEC425.
  • 3. Data protection for AI. AIM373, SEC411 (PII inside AWS Nitro Enclaves), CMP405 (Nitro confidential computing), IND390 (agents that handle PHI) and SEC365.
  • 4. AI governance and compliance. SEC302, SEC386, SEC361, the AIM336-R responsible-AI audit (repeats AIM336-R1), IND353-R (HIPAA-aligned agents), COM401 and the SEC391 AI-supply-chain workshop. For the EU AI Act alongside SOC 2 and PCI DSS, see SEC367 in the  security & compliance post .
Fixed points of the week (official agenda, PT): Expo Welcome Reception Mon 4–7 PM · Expo Tue–Wed 10 AM–6 PM, Expo Happy Hour Wed 4:30–6 PM, Thu 10 AM–4 PM (last chance for swag) · networking receptions Tue 6–8 PM (Caesars Palace, Venetian) and Wed 6–8 PM (Caesars Palace, Venetian, Wynn|Encore) · AMPED Industries & Startup Party Wed 7 PM at AREA15 (RSVP) · re:Play Thu 6:30–10:30 PM at Allegiant Stadium (shuttles from 6:15 PM) · Friday is a half day (limited service at Venetian and Wynn|Encore until ~1 PM). Breakfast and lunch are served on full conference days — eat at the venue of your next session.

Day-by-day (clustered by venue)

Monday Nov 30 — Wynn all day

  • AM · Wynn: the SEC391 Securing your AI-accelerated software supply chain with AWS Well-Architected (8:00 AM, Wynn/Encore) workshop (reserve it), then SEC302 Governance foundations for AI workloads that grow into autonomous agents (10:00 AM, Wynn/Encore).
  • Late AM · Wynn, pick by role: COM401 AI Didn't Wait for Security: Now What? (11:30 AM, Wynn/Encore), IND353-R Agents under oath: Designing agentic AI to support HIPAA compliance on AWS (11:30 AM, Wynn/Encore) or IND390 Deploying AI Agents that handle PHI with Amazon Bedrock Guardrails (11:30 AM, Wynn/Encore) (same slot), then lunch.
  • PM · Wynn: SEC386 Turn autonomous agents into accountable ones (3:00 PM, Wynn/Encore). Hands-on alternative: skip the late-AM pick, eat early and take the AIM336-R A responsible AI challenge: Audit your agentic AI before the regulator does (12:30 PM, Wynn/Encore) workshop. South alternative: the CMP405 Hands-on: Harden your AI agents with Nitro Confidential Computing (3:30 PM, MGM Grand) workshop.
  • 🌙 Walk 10–15 min to the Venetian for Expo Welcome Reception (4:00 PM, The Venetian), then Developer Keynote (7:30 PM, The Venetian).

Tuesday Dec 1 — keynote, then Wynn

  • AM · Venetian: CEO Keynote – Matt Garman (8:30 AM, The Venetian) and the Expo, then walk 10–15 min to Wynn.
  • Midday · Wynn: SEC328-R Enforce AI guardrails across your organization without the friction (11:30 AM, Wynn/Encore), then lunch.
  • PM · Wynn: at 3:30, the SEC407-R Constrain AI agents with permissions boundaries and Cedar in AgentCore (3:30 PM, Wynn/Encore) or SEC312 Use AWS Continuum to move from annual pentesting to continuous security (3:30 PM, Wynn/Encore) workshop (reserve). Walk-up alternative: SEC398 The latest innovations in multicloud agent security (4:30 PM, Wynn/Encore).
  • South alternative: shuttle 30+ min to MGM Grand for INV525 Virtualization Architecture Behind AgentCore's Runtime Security Model (1:30 PM, MGM Grand), then SEC352-R1 Build a security boundary for AI agents with WAF and Network Firewall (4:30 PM, MGM Grand).
  • 🌙 Networking receptions (Caesars Palace & The Venetian) (6:00 PM, The Venetian or Caesars Palace) (from MGM Grand, Caesars Palace is closest).

Wednesday Dec 2 — keynote, then Wynn

  • AM · Venetian: Data & AI Keynote (8:30 AM, The Venetian), then walk to Wynn. Skip-the-keynote alternative: SEC425 From proofs to production: Neurosymbolic AI and provable security (9:00 AM, Wynn/Encore).
  • Late AM · Wynn, pick one: SEC352-R Build a security boundary for AI agents with WAF and Network Firewall (10:00 AM, Wynn/Encore) or PEX401-R1 Building a Zero Trust for Agentic AI Practice: A Partner Framework (10:30 AM, Wynn/Encore), then an early lunch.
  • Midday · Wynn: SEC321-R Securing AI coding agents: defense in depth across the development lifecycle (12:00 PM, Wynn/Encore).
  • PM · Wynn: the AIM336-R1 A responsible AI challenge: Audit your agentic AI before the regulator does (3:00 PM, Wynn/Encore) workshop if you missed it Monday.
  • Center alternative: AIM373 What happens in Bedrock stays in Bedrock (10:00 AM, Caesars Forum), then the SEC407-R1 Constrain AI agents with permissions boundaries and Cedar in AgentCore (3:00 PM, Caesars Forum) workshop, both at Caesars Forum.
  • South alternative: AIM451 Automated Reasoning as Agent Guardrails (12:00 PM, MGM Grand) or the NET321-R1 Protect MCP endpoints on Amazon Bedrock AgentCore Gateway with AWS WAF (12:00 PM, MGM Grand) workshop (same slot).
  • 🌙 Expo Happy Hour (4:30 PM, The Venetian), then Networking receptions (Caesars Palace, The Venetian, Wynn|Encore) (6:00 PM, The Venetian or Caesars Palace / Wynn|Encore) or AMPED: Industries & Startup Party (RSVP in portal) (7:00 PM, AREA15).

Thursday Dec 3 — Wynn AI-security path

Skip the keynote and watch the replay. This path runs alongside the security day in the  security & compliance post  and skips the SEC001 and SEC002 mainstage talks.
  • AM · Wynn, pick a workshop: SEC362 Authentication, authorization, and audit for agentic AI on AWS (8:00 AM, Wynn/Encore) or SEC361 Closing the compliance loop when agents change your infrastructure (8:30 AM, Wynn/Encore). South alternative: AIM453 Securing the agent loop with Amazon Bedrock Guardrails (9:00 AM, MGM Grand).
  • Late AM · Wynn: SEC336 Living Off the Agent: Detecting and Preventing Lateral Movement in Agentic AI (10:30 AM, Wynn/Encore), then SEC382 Mitigating prompt injection with structured outputs in Amazon Bedrock (11:30 AM, Wynn/Encore) or SEC419 OAuth delegation patterns for AI agents with Amazon Bedrock AgentCore Identity (11:30 AM, Wynn/Encore) (same slot).
  • Midday · Wynn: lunch, or SEC343 From MCP poisoning to defense: securing the agent tool supply chain (12:00 PM, Wynn/Encore) first.
  • PM · Wynn: SEC310 Building production-ready agents securely: lessons from AWS at scale (1:00 PM, Wynn/Encore), or SEC327 Securing Multi-Agent Systems (1:30 PM, Wynn/Encore) at 1:30. At 3:00: the SEC341 Build layered defenses for agentic AI from prompt to tool execution (3:00 PM, Wynn/Encore) workshop, the SEC403 Red team CTF: when auth and guardrails aren't enough (3:00 PM, Wynn/Encore) CTF, SEC344 Zero trust for non-human identities: IAM for autonomous agents (3:00 PM, Wynn/Encore) or the SEC003 Your new playbook: Operationalizing security in a world of AI (3:00 PM, Wynn/Encore) mainstage. Or, at 3:30, the SEC409 Threat detection and IR in AgentCore workloads (3:30 PM, Wynn/Encore) or SEC411 Processing PII without exposing it with AWS Nitro Enclaves for confidential computing (3:30 PM, Wynn/Encore) workshop.
  • The Expo closes at 4 PM. 🌙 re:Play (6:30 PM, Allegiant Stadium).

Friday Dec 4 — light, Wynn only

  • INV503-R1 Beyond Benchmarks: Evaluating AI Safety with Formal Guarantees (8:30 AM, Wynn/Encore) or SEC429-R Tenant-isolated agent memory: one IAM policy for AgentCore Memory (9:00 AM, Wynn/Encore) (they overlap), then SEC365 Control what goes out: egress security for AI workloads (11:30 AM, Wynn/Encore). Service ends around 1 PM.
Transit rules of thumb (estimates): Venetian↔Wynn/Encore 10–15 min walk · Venetian↔Caesars Forum 15–20 min · Caesars Forum↔Caesars Palace 5–10 min (0.4 mi — they are different buildings) · Center↔MGM Grand ~20–25 min by shuttle · North↔MGM Grand 30+ min by shuttle. Shuttles run ~6 AM to early evening (check the AWS Events app for live times).

Session table

Code
Title
Type / Level
Day / Time (PT)
Venue
Add to agenda


SEC391
Securing your AI-accelerated software supply chain with AWS Well-Architected
Workshop 300
Mon Nov 30 · 8:00 AM–10:00 AM
Wynn/Encore · Cristal 5
 Catalog ★ 


SEC302
Governance foundations for AI workloads that grow into autonomous agents
Breakout 300
Mon Nov 30 · 10:00 AM–11:00 AM
Wynn/Encore · Cristal 2, Content Hub, Light Blue Theater
 Catalog ★ 


COM401
AI Didn't Wait for Security: Now What?
Breakout 400
Mon Nov 30 · 11:30 AM–12:30 PM
Wynn/Encore · Encore Ballroom 5
 Catalog ★ 


IND353-R
Agents under oath: Designing agentic AI to support HIPAA compliance on AWS
Chalk talk 300
Mon Nov 30 · 11:30 AM–12:30 PM
Wynn/Encore · Chopin 2
 Catalog ★ 


IND390
Deploying AI Agents that handle PHI with Amazon Bedrock Guardrails
Chalk talk 300
Mon Nov 30 · 11:30 AM–12:30 PM
Wynn/Encore · Montrachet 1
 Catalog ★ 


AIM336-R
A responsible AI challenge: Audit your agentic AI before the regulator does
Workshop 300
Mon Nov 30 · 12:30 PM–2:30 PM
Wynn/Encore · Margaux 2
 Catalog ★ 


SEC386
Turn autonomous agents into accountable ones
Chalk talk 300
Mon Nov 30 · 3:00 PM–4:00 PM
Wynn/Encore · Debussy 1
 Catalog ★ 


CMP405
Hands-on: Harden your AI agents with Nitro Confidential Computing
Workshop 400
Mon Nov 30 · 3:30 PM–5:30 PM
MGM Grand · Grand 124
 Catalog ★ 


EXPO-WELCOME
Expo Welcome Reception
Event
Mon Nov 30 · 4:00 PM–7:00 PM
The Venetian · Hall B, Expo
 Catalog ★ 


KEY-DEV
Developer Keynote
Keynote
Mon Nov 30 · 7:30 PM–8:45 PM
The Venetian · Ballroom E, Keynote
 Catalog ★ 


KEY-CEO
CEO Keynote – Matt Garman
Keynote
Tue Dec 1 · 8:30 AM–10:30 AM
The Venetian · Ballroom E, Keynote
 Catalog ★ 


SEC328-R
Enforce AI guardrails across your organization without the friction
Chalk talk 300
Tue Dec 1 · 11:30 AM–12:30 PM
Wynn/Encore · Chopin 2
 Catalog ★ 


INV525
Virtualization Architecture Behind AgentCore's Runtime Security Model
Chalk talk 500
Tue Dec 1 · 1:30 PM–2:30 PM
MGM Grand · Premier 309
 Catalog ★ 


SEC312
Use AWS Continuum to move from annual pentesting to continuous security
Workshop 300
Tue Dec 1 · 3:30 PM–5:30 PM
Wynn/Encore · Margaux 2
 Catalog ★ 


SEC407-R
Constrain AI agents with permissions boundaries and Cedar in AgentCore
Workshop 400
Tue Dec 1 · 3:30 PM–5:30 PM
Wynn/Encore · Cristal 7
 Catalog ★ 


SEC352-R1
Build a security boundary for AI agents with WAF and Network Firewall
Builders' session 300
Tue Dec 1 · 4:30 PM–5:30 PM
MGM Grand · Room 104
 Catalog ★ 


SEC398
The latest innovations in multicloud agent security
Breakout 300
Tue Dec 1 · 4:30 PM–5:30 PM
Wynn/Encore · Cristal 2, Content Hub, Red Theater
 Catalog ★ 


RECEPT-TUE
Networking receptions (Caesars Palace & The Venetian)
Event
Tue Dec 1 · 6:00 PM–8:00 PM
The Venetian or Caesars Palace
 Catalog ★ 


KEY-DATA
Data & AI Keynote
Keynote
Wed Dec 2 · 8:30 AM–10:00 AM
The Venetian · Ballroom E, Keynote
 Catalog ★ 


SEC425
From proofs to production: Neurosymbolic AI and provable security
Breakout 400
Wed Dec 2 · 9:00 AM–10:00 AM
Wynn/Encore · Cristal 2, Content Hub, Red Theater
 Catalog ★ 


AIM373
What happens in Bedrock stays in Bedrock
Breakout 300
Wed Dec 2 · 10:00 AM–11:00 AM
Caesars Forum · Forum 120, Content Hub, Purple Theater
 Catalog ★ 


SEC352-R
Build a security boundary for AI agents with WAF and Network Firewall
Builders' session 300
Wed Dec 2 · 10:00 AM–11:00 AM
Wynn/Encore · Cristal 2, Content Hub, Builders' Session 1
 Catalog ★ 


PEX401-R1
Building a Zero Trust for Agentic AI Practice: A Partner Framework
Builders' session 400
Wed Dec 2 · 10:30 AM–11:30 AM
Wynn/Encore · Latour 7
 Catalog ★ 


AIM451
Automated Reasoning as Agent Guardrails
Chalk talk 400
Wed Dec 2 · 12:00 PM–1:00 PM
MGM Grand · Boulevard 162
 Catalog ★ 


NET321-R1
Protect MCP endpoints on Amazon Bedrock AgentCore Gateway with AWS WAF
Workshop 300
Wed Dec 2 · 12:00 PM–2:00 PM
MGM Grand · Premier 314
 Catalog ★ 


SEC321-R
Securing AI coding agents: defense in depth across the development lifecycle
Chalk talk 300
Wed Dec 2 · 12:00 PM–1:00 PM
Wynn/Encore · Debussy 1
 Catalog ★ 


AIM336-R1
A responsible AI challenge: Audit your agentic AI before the regulator does
Workshop 300
Wed Dec 2 · 3:00 PM–5:00 PM
Wynn/Encore · Brahms 4
 Catalog ★ 


SEC407-R1
Constrain AI agents with permissions boundaries and Cedar in AgentCore
Workshop 400
Wed Dec 2 · 3:00 PM–5:00 PM
Caesars Forum · Alliance 307
 Catalog ★ 


EXPO-HH
Expo Happy Hour
Event
Wed Dec 2 · 4:30 PM–6:00 PM
The Venetian · Expo
 Catalog ★ 


RECEPT-WED
Networking receptions (Caesars Palace, The Venetian, Wynn
Encore)
Event
Wed Dec 2 · 6:00 PM–8:00 PM
The Venetian or Caesars Palace / Wynn
Encore
 Catalog ★ 
AMPED
AMPED: Industries & Startup Party (RSVP in portal) (end time not published; RSVP required)
Event
Wed Dec 2 · 7:00 PM
AREA15
 Catalog ★ 


SEC362
Authentication, authorization, and audit for agentic AI on AWS
Workshop 300
Thu Dec 3 · 8:00 AM–10:00 AM
Wynn/Encore · Cristal 5
 Catalog ★ 


SEC361
Closing the compliance loop when agents change your infrastructure
Workshop 300
Thu Dec 3 · 8:30 AM–10:30 AM
Wynn/Encore · Cristal 7
 Catalog ★ 


AIM453
Securing the agent loop with Amazon Bedrock Guardrails
Chalk talk 400
Thu Dec 3 · 9:00 AM–10:00 AM
MGM Grand · Boulevard 165
 Catalog ★ 


SEC336
Living Off the Agent: Detecting and Preventing Lateral Movement in Agentic AI
Chalk talk 300
Thu Dec 3 · 10:30 AM–11:30 AM
Wynn/Encore · La Tache 2
 Catalog ★ 


SEC382
Mitigating prompt injection with structured outputs in Amazon Bedrock
Code talk 300
Thu Dec 3 · 11:30 AM–12:30 PM
Wynn/Encore · Cristal 2, Content Hub, Code Talk
 Catalog ★ 


SEC419
OAuth delegation patterns for AI agents with Amazon Bedrock AgentCore Identity
Chalk talk 400
Thu Dec 3 · 11:30 AM–12:30 PM
Wynn/Encore · Castillon
 Catalog ★ 


SEC343
From MCP poisoning to defense: securing the agent tool supply chain
Builders' session 300
Thu Dec 3 · 12:00 PM–1:00 PM
Wynn/Encore · Cristal 2, Content Hub, Builders' Session 2
 Catalog ★ 


SEC310
Building production-ready agents securely: lessons from AWS at scale
Code talk 300
Thu Dec 3 · 1:00 PM–2:00 PM
Wynn/Encore · Cristal 2, Content Hub, Code Talk
 Catalog ★ 


SEC327
Securing Multi-Agent Systems
Chalk talk 300
Thu Dec 3 · 1:30 PM–2:30 PM
Wynn/Encore · Fleurie
 Catalog ★ 


SEC003
Your new playbook: Operationalizing security in a world of AI
Mainstage
Thu Dec 3 · 3:00 PM–4:00 PM
Wynn/Encore · Bandol 1
 Catalog ★ 


SEC341
Build layered defenses for agentic AI from prompt to tool execution
Workshop 300
Thu Dec 3 · 3:00 PM–5:00 PM
Wynn/Encore · Cristal 1
 Catalog ★ 


SEC344
Zero trust for non-human identities: IAM for autonomous agents
Builders' session 300
Thu Dec 3 · 3:00 PM–4:00 PM
Wynn/Encore · Cristal 2, Content Hub, Builders' Session 2
 Catalog ★ 


SEC403
Red team CTF: when auth and guardrails aren't enough
Workshop 400
Thu Dec 3 · 3:00 PM–5:00 PM
Wynn/Encore · Cristal 5
 Catalog ★ 


SEC409
Threat detection and IR in AgentCore workloads
Workshop 400
Thu Dec 3 · 3:30 PM–5:30 PM
Wynn/Encore · Cristal 3
 Catalog ★ 


SEC411
Processing PII without exposing it with AWS Nitro Enclaves for confidential computing
Workshop 400
Thu Dec 3 · 3:30 PM–5:30 PM
Wynn/Encore · Cristal 7
 Catalog ★ 


REPLAY
re:Play
Event
Thu Dec 3 · 6:30 PM–10:30 PM
Allegiant Stadium
 Catalog ★ 


INV503-R1
Beyond Benchmarks: Evaluating AI Safety with Formal Guarantees
Chalk talk 500
Fri Dec 4 · 8:30 AM–9:30 AM
Wynn/Encore · Chopin 2
 Catalog ★ 


SEC429-R
Tenant-isolated agent memory: one IAM policy for AgentCore Memory
Chalk talk 400
Fri Dec 4 · 9:00 AM–10:00 AM
Wynn/Encore · La Tache 2
 Catalog ★ 


SEC365
Control what goes out: egress security for AI workloads
Breakout 300
Fri Dec 4 · 11:30 AM–12:30 PM
Wynn/Encore · Cristal 2, Content Hub, Light Blue Theater
 Catalog ★ 


* = estimated end time · Keynotes and breakouts are walk-up, so arrive 15 minutes early · Confirm rooms in the AWS Events app.
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article