Runtime authorization for telecom AI agents.
A carrier-care or B2B-telco agent with OSS/BSS, NumberPortal, and HSS credentials can authorize a SIM swap, port a number out to a fraudster, change a subscriber's plan without consent, or fire a million-text campaign in a single tool loop. SSO and OAuth confirm the agent reached the API. Veto decides whether the specific port, SIM provisioning, plan change, or outbound message is allowed under your FCC, CPNI, and TCPA rules.
What can go wrong
Telecom accounts are high-value targets: the phone number often becomes the credential for SMS-based MFA across the financial system. Port-out, SIM-change, and customer-service workflows need explicit verification gates before an agent can touch subscriber identity.
SIM-swap and port-out fraud
A customer-care agent is socially engineered into porting a subscriber's number to a fraudster's device. From there, the attacker has the SMS-based 2FA for every bank and exchange the victim uses.
Plan and feature changes without consent
A retention agent reads "the customer wants to switch plans" out of an inbound chat from a bot, and commits a downgrade that triggers a $200 early-termination charge. The customer sees the charge before support catches it.
Mass-text TCPA violations
A growth agent sends a "win-back" SMS to a CRM segment that includes opted-out and DNC numbers. The send should fail before the campaign leaves the queue.
CALEA and lawful-intercept exposure
An ops agent tasked with "checking line provisioning" enumerates tools that touch lawful-intercept configuration. Even a read of these systems by a non-authorized identity is a CALEA compliance event.
What Veto enforces
Wrap the OSS/BSS, number-portability, provisioning, and outbound-messaging tools your agent uses. Argument-level policies check verification, channel, time, and audience size.
policies:
- name: "Port-out and SIM-swap protection"
match:
tool: ["bss.initiate_port_out","provisioning.swap_sim","provisioning.add_esim"]
rules:
- condition: "subscriber.mfa_verified != true"
action: deny
reason: "FCC 47 CFR 64.2010 verification required"
- condition: "subscriber.recent_device_change_hours < 24"
action: require_approval
approvers: ["fraud-ops@carrier.invalid"]
- condition: "request.origin_state != line.state_of_record"
action: require_approval
reason: "Out-of-state port request"
- condition: "line.high_risk_score > 0.7"
action: deny
reason: "Fraud-score threshold tripped"
- name: "Plan and feature change consent"
match:
tool: ["bss.change_plan","bss.add_feature","bss.remove_line"]
rules:
- condition: "args.triggers_etf == true && args.consent_method != 'recorded_voice'"
action: deny
reason: "ETF-triggering change requires recorded consent"
- condition: "tool == 'bss.remove_line'"
action: require_approval
approvers: ["account-holder","retention-supervisor@carrier.invalid"]
- condition: "args.affected_lines > 5"
action: require_approval
- name: "TCPA outbound-SMS guard"
match:
tool: ["sms.send_bulk","mms.send_bulk"]
rules:
- condition: "any(recipients, r => r.dnc == true || r.opted_out == true)"
action: deny
reason: "TCPA and DNC violation"
- condition: "recipient_local_time outside ['08:00','21:00']"
action: deny
reason: "TCPA quiet-hours rule"
- condition: "recipients.length > 1000"
action: require_approval
approvers: ["growth-lead@carrier.invalid","compliance@carrier.invalid"]
- name: "CALEA isolation"
match:
tool: ["lawful_intercept.*","wiretap.*"]
rules:
- condition: "true"
action: deny
reason: "CALEA-relevant tools are out of agent scope"Evidence hooks
CPNI (47 CFR 64.2010)
Carriers must authenticate the customer before disclosing or modifying account data. Veto's MFA-verified check is the artifact the FCC asks for after a breach.
FCC port-out and SIM-change rules (2024)
Carriers verify the subscriber and notify on old and new lines. Veto per-call policy records can preserve the evidence record for that verification.
CALEA
Lawful-intercept paths are off-limits for AI customer-service agents. A categorical deny rule can enforce this at the tool-call layer.
TCPA
DNC, opt-out, consent-source, and quiet-hours rules can be checked at the send tool before the message goes out.
Where controls matter
Telecom regulators increasingly expect stronger authentication, access controls, and record-keeping across customer-care workflows, including AI-assisted ones.
Port-out, SIM-change, and outbound messaging tools need explicit verification, opt-out, quiet-hours, and high-volume approval rules before the action reaches a carrier system.
Frequently asked questions
How does Veto reduce SIM-swap and port-out fraud risk?
Does this work with CALEA lawful-intercept obligations?
Can agents still process routine plan and feature changes?
How does this map to TCPA and FCC mass-text rules?
Related use cases
Stop one SIM swap. Save a customer and a federal investigation.
Pick one OSS/BSS, port, or messaging policy.